Career Employer

Your FREE CompTIA Cloud+ (CV0-004) Practice Test 2026 – 190+ Q&A

Realistic CompTIA Cloud+ CV0-004 practice questions across all six official exam domains, with instant scoring and answer explanations.

How ready are you?

To find us again, just search “Career Employer Cloud+”

By
Cloud+ exam by the numbersStandard mode

Where the Cloud+ exam's points come from

CompTIA's official outline weights, the exam rules and our 199 free practice questions mapped domain by domain. Use it to plan your study hours.

  • Certifying Body:CompTIA
  • Exam Code:CV0-004 (current version)
  • Total Questions:Maximum of 90 (multiple choice and performance-based)
  • Exam Cost:About USD $390 (subject to change; verify with CompTIA)
  • Recertification:Valid 3 years; renew via CompTIA continuing education (CEUs)
Facts from CompTIA's official exam outline and rules
90 min
Time Limit
750
Passing Score
2-3 years
Recommended Experience

Where the Cloud+ points are

CompTIA's official weight for each of the six domains. Hover or tap a domain to see how many questions that means on your exam.

See these numbers as a table
Weights from CompTIA's Cloud+ exam outline. Questions on your exam = weight × 90 questions, rounded. Practice questions: Career Employer's Cloud+ bank.
DomainOfficial weightQuestions on your examOur practice questions
Cloud Architecture23%2146
Deployment19%1741
Security19%1737
Operations17%1533
Troubleshooting12%1123
DevOps Fundamentals10%919

What to do: Cloud Architecture is the biggest domain at 23%, which means 21 of your questions. DevOps Fundamentals is the smallest at 10%.

Split your study time

Your hours, divided by CompTIA's domain weights.

Cloud Architecture
14 hrs
Deployment
11 hrs
Security
11 hrs
Operations
10 hrs
Troubleshooting
7.2 hrs
DevOps Fundamentals
6.0 hrs

Our Cloud+ student data is on its way

We’ve logged 2,661 first-try answers on our Cloud+ questions so far. This section switches to real Cloud+ student results by itself once each part has enough people behind it:

What switches onNeeds
Headline numbers150 studiers
Accuracy by domain300 answers + 50 people per domain
First full exam score100 full exams
Finish rate150 people starting

Then it shows points lost by domain, you vs everyone and question difficulty.

Weights and rules from CompTIA's published Cloud+ exam outline and certification pages. Practice-question counts from Career Employer's Cloud+ bank. Fees and requirements can change; check CompTIA before you register.

CompTIA Cloud+, exam code CV0-004, is a vendor-neutral certification that validates the skills needed to deploy and automate secure cloud environments supporting the high availability of modern applications and data.[1] These free Cloud+ practice questions and test prep mirror the official CV0-004 objectives so you practice the way the real exam is built.[2] To round out your prep, pair these with our free study guide, flashcards, and cheat sheet.

CompTIA Cloud+ is one of the 14 CompTIA certifications — explore all our CompTIA practice tests to compare and prep across the whole family.

What Is on the CompTIA Cloud+ Exam?

The CompTIA Cloud+ CV0-004 exam covers six domains: Cloud Architecture (23%), Deployment (19%), Security (19%), Operations (17%), Troubleshooting (12%), and DevOps Fundamentals (10%).[1]

Cloud Architecture is the largest section, covering cloud and deployment models, capacity planning, and high availability. CV0-004 is the first CompTIA certification to formally include a DevOps domain. Our full practice test is weighted to match.

CompTIA Cloud+ practice test — practice questions by domain with answer explanations

Practice Questions by Domain

Use Start Test for a full weighted CV0-004 simulation, or open the hub and pick a single domain to drill your weak area. After each full exam, your results show a per-domain breakdown so you know exactly where to focus — most candidates need the most reps on Cloud Architecture and the new DevOps content.

What Are the Requirements to Take CompTIA Cloud+?

CompTIA Cloud+ has no mandatory prerequisites — the exam is open to anyone who registers and pays the fee.[1] CompTIA recommends that candidates have 2 to 3 years of hands-on experience as a systems administrator or cloud engineer, along with familiarity with a major cloud service provider, before attempting CV0-004. Network+ and Server+ (or equivalent knowledge) provide useful foundational background.

How Do You Register for the CompTIA Cloud+ Exam?

You register for CompTIA Cloud+ (CV0-004) by purchasing an exam voucher from the CompTIA Marketplace (store.comptia.org) or an authorized reseller, then scheduling your exam through Pearson VUE.[5] You can test at a Pearson VUE testing center or via online proctored delivery from home.

After purchasing a voucher you create or sign in to your CompTIA and Pearson VUE accounts to pick a date, time, and delivery method. Verify the current voucher price and any bundle or retake options on the CompTIA Marketplace, as prices vary by region and promotion.

What Is the Passing Score for CompTIA Cloud+?

The passing score for CompTIA Cloud+ CV0-004 is 750 on a scale of 100 to 900.[1]

Scores are not a simple percentage of questions correct; CompTIA uses a scaled scoring model so questions of differing difficulty contribute appropriately, and performance-based questions are weighted alongside multiple-choice items.

You receive a pass or fail result immediately at the testing center, along with a score report that breaks down your performance by domain. Your overall scaled score — not any single domain — determines whether you pass.

How Hard Is Cloud+? (Pass Rate)

Cloud+ is an intermediate-level certification — more advanced than entry-level certs like A+ or Network+, but not as specialized as expert cloud credentials. CompTIA does not publish official pass-rate statistics, so any specific figure cited online is an unofficial estimate.[4] The difficulty comes from its vendor-neutral breadth and the performance-based questions; candidates most often underestimate the new DevOps and security content.

Intermediate
Difficulty level
above A+/Network+
23%
Cloud Architecture
largest domain

The takeaway: practice automation, security hardening, and troubleshooting workflows across providers, and drill until you’re consistently scoring above target on full-length practice before you book your exam date.

What to Expect on Exam Day

Arrive at your Pearson VUE test center at least 15 minutes early to check in — bring a valid, unexpired government-issued photo ID whose name matches your CompTIA registration.[5]You’ll store phones and personal items in a locker; no notes are allowed.

A short tutorial precedes the exam, then you have 90 minutes to answer up to 90 multiple-choice and performance-based questions. The PBQs ask you to complete tasks in a simulated environment, so budget time carefully — they take longer than standard items.

If you test via online proctoring, expect a workspace scan and ID check. CompTIA reports a pass/fail result immediately, with a domain breakdown on your score report. Having simulated the full timing with practice tests makes that 90-minute clock feel routine.

How to Use This Cloud+ Practice Test

  • Recreate exam conditions. Take the full test timed, with no notes.[4]
  • Diagnose, then drill. Use a full CV0-004 simulation to find weak domains, then drill them.
  • Prioritize Architecture + DevOps. Architecture is the largest domain and DevOps is the newest.
  • Learn the why. Read every explanation — understanding beats memorizing.
  • Think vendor-neutral. Practice concepts across providers, not just one platform.

Why Get Cloud+ Certified?

Cloud+ is a vendor-neutral credential that validates multi-cloud deployment, security, and automation skills employers value for cloud engineer, systems administrator, and cloud security roles, and CV0-004 keeps it current with modern DevOps practices.[1] These free Cloud+ practice tests are the most efficient way to get there.

Conclusion

Passing CV0-004 comes down to knowing cloud architecture, security, automation, and the new DevOps fundamentals cold. Use this free CompTIA Cloud+ practice test to find your weak domains, drill them to mastery, and reinforce them with our study guide, flashcards, and cheat sheet. Start with one full-length practice test to find your weakest section, then drill that section first.

CompTIA Cloud+ Practice Test FAQ

The current version is CV0-004, which launched on September 24, 2024 and replaced the previous CV0-003. CV0-004 is the first CompTIA certification to include a dedicated DevOps domain. When you register, make sure you are scheduling the CV0-004 exam.

What changed on this page

Every real change, with the date.

  1. Interactive charts added. Built from the official outline and our question bank.
  2. 1 question corrected.
  3. 199 questions rewritten.
  4. 67 new practice questions. 6 questions retired. The bank now has 199 questions.

Cloud+ question bank

All 199 questions, by domain

A reference copy of every question in this practice test. Each answer stays hidden until you choose to show it. To practice with scoring, timing and your readiness score, use Start Test at the top of the page.

Cloud Architecture (46)

  1. What is the primary benefit of implementing a multi-cloud strategy?

    • A.Retention of on-premises data
    • B.Consolidation of provider billing
    • C.Prevention of vendor lock-in
    • D.Elasticity of compute capacity
    Show answerHide answer

    Correct answer: Prevention of vendor lock-in

    Prevention of vendor lock-in is the main benefit of a multi-cloud strategy: spreading workloads across two or more providers keeps the organization from depending on any single vendor's services, prices and terms. Retention of on-premises data is the reason to choose a hybrid cloud, which pairs a private environment with a public one. Consolidation of provider billing is an advantage of staying with one provider; multi-cloud multiplies the bills to manage. Elasticity of compute capacity comes with any public cloud and does not depend on using several providers.

  2. Which technology is essential for creating a hybrid cloud environment?

    • A.Virtual private network (VPN)
    • B.Virtual local area network (VLAN)
    • C.Storage area network (SAN)
    • D.Content delivery network (CDN)
    Show answerHide answer

    Correct answer: Virtual private network (VPN)

    A virtual private network (VPN) is the essential link in a hybrid cloud: an encrypted tunnel joins the private data center to the public cloud so workloads and data can move securely between them. A virtual local area network (VLAN) only segments traffic inside one network and does not join two sites. A storage area network (SAN) gives servers block-level access to shared disks. A content delivery network (CDN) caches content near users rather than connecting private and public environments.

  3. Which cloud service model delivers software applications over the internet and is managed by a third-party vendor?

    • A.Infrastructure as a Service (IaaS)
    • B.Platform as a Service (PaaS)
    • C.Software as a Service (SaaS)
    • D.Database as a Service (DBaaS)
    Show answerHide answer

    Correct answer: Software as a Service (SaaS)

    Software as a Service (SaaS) delivers complete applications over the internet, with the vendor running the infrastructure, platform and software; users simply sign in and use them. Infrastructure as a Service (IaaS) rents virtual machines, storage and networks that the customer still manages. Platform as a Service (PaaS) supplies a managed runtime on which customers build and run their own applications. Database as a Service (DBaaS) provides a managed database engine, not finished end-user applications.

  4. In cloud computing, what is the primary function of a load balancer?

    • A.To cache static web content across edge locations
    • B.To add compute instances across availability zones
    • C.To spread client requests across multiple servers
    • D.To replicate database records across separate regions
    Show answerHide answer

    Correct answer: To spread client requests across multiple servers

    To spread client requests across multiple servers is the function of a load balancer: it distributes network or application traffic so no single server is overwhelmed, which improves throughput and availability. To cache static web content across edge locations is the work of a content delivery network (CDN). To add compute instances across availability zones is auto-scaling. To replicate database records across separate regions is data replication for durability and disaster recovery.

  5. What is the primary purpose of auto-scaling in a cloud environment?

    • A.To steer incoming user traffic to healthy instances
    • B.To adjust compute capacity to match current demand
    • C.To shift failed workloads to standby recovery regions
    • D.To shrink oversized instances to reduce monthly costs
    Show answerHide answer

    Correct answer: To adjust compute capacity to match current demand

    To adjust compute capacity to match current demand is the purpose of auto-scaling: it adds instances when load rises and removes them when it falls, keeping performance steady without paying for idle servers. To steer incoming user traffic to healthy instances is load balancing, which spreads requests but does not change capacity. To shift failed workloads to standby recovery regions is failover for disaster recovery. To shrink oversized instances to reduce monthly costs is right-sizing, a periodic change of instance size rather than scaling with load.

  6. Which cloud computing feature allows for the automatic distribution of incoming application traffic across multiple targets, such as EC2 instances?

    • A.Elastic Container Service
    • B.Elastic Load Balancing
    • C.Amazon EC2 Auto Scaling
    • D.Amazon CloudFront
    Show answerHide answer

    Correct answer: Elastic Load Balancing

    Elastic Load Balancing automatically distributes incoming application traffic across multiple targets such as EC2 instances, containers and IP addresses, sending requests only to healthy targets. Elastic Container Service runs and schedules containers; it can register them with a load balancer but does not spread the traffic itself. Amazon EC2 Auto Scaling adds or removes instances as demand changes, which is a different job from dividing requests among them. Amazon CloudFront is a content delivery network that caches content at edge locations.

  7. Which cloud deployment model is best suited for organizations with fluctuating resource demands, offering a scalable and flexible environment?

    • A.Private cloud
    • B.Community cloud
    • C.Hosted private cloud
    • D.Public cloud
    Show answerHide answer

    Correct answer: Public cloud

    Correct answer: Public cloud. Explanation: Public cloud suits fluctuating demand best: the provider's large shared pool lets an organization scale up or down on demand and pay only for what it uses. A private cloud is limited to the capacity the organization has built or reserved for itself. A hosted private cloud runs at a provider but is still dedicated to one organization, so its capacity is fixed by what was provisioned. A community cloud is shared by organizations with common requirements, and its capacity is still bounded by that group.

  8. What is the primary advantage of using serverless computing in a cloud environment?

    • A.Customer control of the operating system
    • B.Fixed monthly pricing for reserved capacity
    • C.Provider management of the underlying servers
    • D.Dedicated physical hardware for isolated workloads
    Show answerHide answer

    Correct answer: Provider management of the underlying servers

    Provider management of the underlying servers is the main advantage of serverless computing: the cloud provider provisions, patches and scales the underlying servers, so developers only deploy code. Customer control of the operating system is an advantage of infrastructure as a service, and serverless gives it up. Fixed monthly pricing for reserved capacity belongs to reserved instances; serverless is billed per execution. Dedicated physical hardware for isolated workloads describes dedicated hosts, whereas serverless functions run on shared provider infrastructure.

  9. In cloud computing, what is the purpose of a virtual network function (VNF)?

    • A.To separate network control from packet forwarding hardware
    • B.To segregate network devices into isolated broadcast domains
    • C.To poll physical network devices for health statistics
    • D.To virtualize network appliances on standard server hardware
    Show answerHide answer

    Correct answer: To virtualize network appliances on standard server hardware

    To virtualize network appliances on standard server hardware is the purpose of a virtual network function (VNF): routing, firewall and load-balancing functions that once needed dedicated hardware run as software on standard servers under network function virtualization. To separate network control from packet forwarding hardware describes software-defined networking (SDN), a related but different idea. To segregate network devices into isolated broadcast domains is what VLANs do. To poll physical network devices for health statistics is network monitoring, for example with SNMP.

  10. What does the term "cloud bursting" refer to in cloud computing?

    • A.Failing over to a standby cloud region after an outage
    • B.Overflowing into public cloud capacity when demand peaks
    • C.Losing cloud performance to a busy neighboring tenant
    • D.Leaving unused cloud instances active after projects end
    Show answerHide answer

    Correct answer: Overflowing into public cloud capacity when demand peaks

    Overflowing into public cloud capacity when demand peaks is cloud bursting: an application normally runs in a private cloud or data center and spills the extra load into a public cloud during spikes, then scales back. Failing over to a standby cloud region after an outage is disaster recovery failover, triggered by failure rather than demand. Losing cloud performance to a busy neighboring tenant is the noisy-neighbor problem of shared infrastructure. Leaving unused cloud instances active after projects end is cloud sprawl, which wastes money.

  11. Which aspect of cloud computing allows for the division of a single physical server into multiple virtual machines?

    • A.Containerization
    • B.Clustering
    • C.Multitenancy
    • D.Virtualization
    Show answerHide answer

    Correct answer: Virtualization

    Virtualization uses a hypervisor to divide one physical server into multiple isolated virtual machines, each with its own operating system. Containerization isolates applications that share a single host kernel, so it does not create separate virtual machines. Clustering works the other way, joining several servers so they act as one system. Multitenancy lets many customers share the same infrastructure, and it depends on virtualization rather than performing the division itself.

  12. In the context of cloud computing, what is the main purpose of implementing an API gateway?

    • A.To store API keys and secrets in an encrypted vault
    • B.To generate API reference documentation from source code
    • C.To run automated API tests before each release
    • D.To route client API calls through one entry point
    Show answerHide answer

    Correct answer: To route client API calls through one entry point

    To route client API calls through one entry point is the purpose of an API gateway: every request enters through it, and it handles routing, authentication, rate limiting and monitoring before passing calls to the right back-end service. To store API keys and secrets in an encrypted vault is the job of a secrets manager. To generate API reference documentation from source code is done by documentation tools built on specifications such as OpenAPI. To run automated API tests before each release belongs to the CI/CD pipeline, not the gateway.

  13. What is the primary role of a cloud service broker?

    • A.To audit cloud services for their security and compliance
    • B.To provide a network link between consumers and providers
    • C.To act as an intermediary between consumers and providers
    • D.To own and operate the data centers behind cloud services
    Show answerHide answer

    Correct answer: To act as an intermediary between consumers and providers

    To act as an intermediary between consumers and providers is the role of a cloud service broker, which manages the use, performance and delivery of cloud services and negotiates relationships between the two sides. To audit cloud services for their security and compliance is the role of a cloud auditor. To provide a network link between consumers and providers is the role of a cloud carrier, which supplies connectivity rather than managing the services. To own and operate the data centers behind cloud services is the role of the cloud provider itself.

  14. Which cloud computing service model provides users with a platform to develop, run, and manage applications without the complexity of building and maintaining the underlying infrastructure?

    • A.Infrastructure as a Service (IaaS)
    • B.Software as a Service (SaaS)
    • C.Platform as a Service (PaaS)
    • D.Storage as a Service (STaaS)
    Show answerHide answer

    Correct answer: Platform as a Service (PaaS)

    Platform as a Service (PaaS) gives developers a managed platform, including the runtime, middleware and operating system, on which to build, run and manage applications without maintaining the underlying servers. Infrastructure as a Service (IaaS) leaves the customer to manage operating systems and runtimes on rented virtual machines. Software as a Service (SaaS) delivers finished applications for end users rather than a development platform. Storage as a Service (STaaS) rents storage capacity only.

  15. In a cloud environment, which of the following best describes the concept of 'elasticity'?

    • A.Adding servers and storage for long-term growth
    • B.Growing or shrinking resources as demand changes
    • C.Keeping servers running through hardware failures
    • D.Sharing a pooled set of resources among tenants
    Show answerHide answer

    Correct answer: Growing or shrinking resources as demand changes

    Growing or shrinking resources as demand changes is elasticity: the cloud adds capacity when load rises and releases it when load falls, often automatically. Adding servers and storage for long-term growth is scalability, which handles steady growth but not necessarily shrinking back. Keeping servers running through hardware failures is high availability. Sharing a pooled set of resources among tenants is resource pooling and multitenancy.

  16. In cloud computing, what is the primary purpose of resource pooling?

    • A.To share common capacity among multiple tenants
    • B.To reserve dedicated capacity for individual tenants
    • C.To meter tenant usage for pay-per-use billing
    • D.To scale tenant workloads quickly as demand changes
    Show answerHide answer

    Correct answer: To share common capacity among multiple tenants

    To share common capacity among multiple tenants is the purpose of resource pooling: the provider's compute, storage and network resources form one pool that is assigned and reassigned to many customers as their demand changes. To reserve dedicated capacity for individual tenants is the opposite, single-tenant arrangement. To meter tenant usage for pay-per-use billing is the measured-service characteristic. To scale tenant workloads quickly as demand changes is rapid elasticity, which pooling makes possible but which is a separate characteristic.

  17. Which cloud security strategy involves replicating data across multiple locations to ensure its availability and durability?

    • A.Geographic redundancy
    • B.Local disk mirroring
    • C.Automated storage tiering
    • D.Point-in-time snapshots
    Show answerHide answer

    Correct answer: Geographic redundancy

    Geographic redundancy, or geo-redundancy, keeps replicated copies of data in data centers in different regions, so the data stays available and durable even if a whole site is lost. Local disk mirroring copies data to a second disk in the same system, which protects against a disk failure but not a site disaster. Automated storage tiering moves data between faster and cheaper storage classes according to how often it is used. Point-in-time snapshots capture a volume's state for rollback, usually in the same region as the original.

  18. In cloud computing, what is the primary function of a hypervisor?

    • A.To back up virtual machines to object storage
    • B.To patch virtual machines as updates are released
    • C.To run virtual machines on physical hardware
    • D.To monitor virtual machines for performance alerts
    Show answerHide answer

    Correct answer: To run virtual machines on physical hardware

    A hypervisor's core function is to run virtual machines on physical hardware: it creates each VM and shares the host's CPU, memory and storage among them. 'To back up virtual machines to object storage' is the work of backup software; 'to patch virtual machines as updates are released' is patch management inside each guest; 'to monitor virtual machines for performance alerts' is a monitoring tool's job.

  19. In a cloud computing environment, what does the term 'autoscaling' refer to?

    • A.Matching the number of servers to current demand
    • B.Sending traffic evenly to a fixed group of hosts
    • C.Applying software updates to each host at once
    • D.Moving workloads from one region to another during an outage
    Show answerHide answer

    Correct answer: Matching the number of servers to current demand

    Autoscaling means matching the number of servers to current demand: instances are added as load rises and removed as it falls. Sending traffic evenly to a fixed group of hosts is load balancing, which never changes how many hosts exist; applying software updates to each host at once is patch deployment; moving workloads from one region to another during an outage is failover for disaster recovery.

  20. Which of these is a key feature of cloud-based disaster recovery planning?

    • A.Shipping backup tapes to an offsite vault
    • B.Keeping an idle second data center on standby
    • C.Rebuilding servers by hand from printed runbooks
    • D.Launching replacement resources in minutes
    Show answerHide answer

    Correct answer: Launching replacement resources in minutes

    Cloud-based disaster recovery is built on launching replacement resources in minutes: servers, storage and networks are started from templates and images when a disaster strikes. Shipping backup tapes to an offsite vault, keeping an idle second data center on standby and rebuilding servers by hand from printed runbooks are traditional approaches that the cloud model replaces because they are slow or costly.

  21. What does 'multi-tenancy' in cloud computing refer to?

    • A.A single company using services from several providers
    • B.A single application running in several regions
    • C.Many customers sharing the same infrastructure
    • D.Many workloads moving between private and public clouds
    Show answerHide answer

    Correct answer: Many customers sharing the same infrastructure

    Multi-tenancy means many customers sharing the same infrastructure, with each tenant's data and workloads kept logically separate. A single company using services from several providers is multi-cloud; a single application running in several regions is a multi-region deployment; many workloads moving between private and public clouds describes a hybrid cloud. None of these is about tenants sharing resources.

  22. What is the primary benefit of implementing 'cloud federation' in cloud management?

    • A.Buying capacity in bulk at discounted rates
    • B.Sharing resources smoothly across providers
    • C.Copying data to several availability zones
    • D.Running each workload on dedicated hardware
    Show answerHide answer

    Correct answer: Sharing resources smoothly across providers

    Cloud federation links separate cloud environments so organizations gain from sharing resources smoothly across providers, using common agreements and identity. Buying capacity in bulk at discounted rates is a purchasing model such as reserved instances; copying data to several availability zones is replication within one provider for resilience; running each workload on dedicated hardware describes dedicated hosts, the opposite of sharing.

  23. What is the purpose of a 'cloud broker' in cloud computing environments?

    • A.To negotiate service contracts on behalf of customers
    • B.To audit security controls applied by cloud providers
    • C.To shorten network routes to speed data transfer
    • D.To add another layer of encryption to stored data
    Show answerHide answer

    Correct answer: To negotiate service contracts on behalf of customers

    A cloud broker's purpose is to negotiate service contracts on behalf of customers, acting as an intermediary that selects, combines and manages services on the customer's behalf (NIST SP 500-292). 'To audit security controls applied by cloud providers' is the separate role of a cloud auditor; 'to shorten network routes to speed data transfer' is network optimization such as a CDN; 'to add another layer of encryption to stored data' is a security control, not brokerage.

  24. A finance team adopts a hosted email and document-collaboration suite where the provider manages the application, runtime, operating system, and all underlying hardware, and users simply log in through a browser. Which cloud service model is this?

    • A.Software as a Service (SaaS)
    • B.Function as a Service (FaaS)
    • C.Storage as a Service (STaaS)
    • D.Platform as a Service (PaaS)
    Show answerHide answer

    Correct answer: Software as a Service (SaaS)

    Software as a Service (SaaS) delivers a finished application over the internet, with the provider managing the application, runtime, operating system and hardware, so users just sign in through a browser, exactly as with a hosted email and document suite. Function as a Service (FaaS) runs a developer's own event-triggered code and gives end users no application to log in to. Storage as a Service (STaaS) rents capacity for storing data but does not supply the email and collaboration application. Platform as a Service (PaaS) provides a runtime for deploying the customer's own code, not a ready-made application.

  25. A developer needs an environment that provides a managed runtime, database, and deployment tooling so they can publish a web application without provisioning or patching servers or operating systems. Which cloud service model best fits this need?

    • A.Software as a Service (SaaS)
    • B.Function as a Service (FaaS)
    • C.Database as a Service (DBaaS)
    • D.Platform as a Service (PaaS)
    Show answerHide answer

    Correct answer: Platform as a Service (PaaS)

    Platform as a Service (PaaS) gives developers a managed platform, with runtime, middleware, databases and deployment tooling, on which to publish their own application without provisioning or patching servers or operating systems. Software as a Service (SaaS) delivers someone else's finished application, so there is nowhere to deploy the developer's own code. Function as a Service (FaaS) runs individual event-triggered functions rather than hosting a full web application with its database and deployment tooling. Database as a Service (DBaaS) supplies only the managed database, not a runtime or deployment tooling.

  26. A cloud engineer provisions virtual machines and configures the guest operating systems, runtime, and applications while the provider manages only the physical servers, storage, virtualization layer, and networking. Which service model is in use?

    • A.Platform as a Service (PaaS)
    • B.Disaster Recovery as a Service (DRaaS)
    • C.Infrastructure as a Service (IaaS)
    • D.Software as a Service (SaaS)
    Show answerHide answer

    Correct answer: Infrastructure as a Service (IaaS)

    Infrastructure as a Service (IaaS) supplies virtualized compute, storage and networking on provider-managed hardware, while the customer installs and manages the guest operating system, runtime and applications, which is exactly the split described. Platform as a Service (PaaS) would also take the operating system and runtime off the customer's hands. Disaster Recovery as a Service (DRaaS) replicates systems to a provider so they can fail over after an outage; it is a recovery service, not the model for building and running everyday virtual machines. Software as a Service (SaaS) delivers a finished application and leaves the customer nothing in the stack to configure.

  27. A team must distinguish three cloud service models for a procurement decision. Which statement correctly describes the difference between IaaS, PaaS, and SaaS?

    • A.IaaS provides complete software applications, PaaS provides virtual server infrastructure, and SaaS provides managed runtime platforms
    • B.IaaS provides managed runtime platforms, PaaS provides complete software applications, and SaaS provides virtual server infrastructure
    • C.IaaS provides virtual server infrastructure, PaaS provides managed runtime platforms, and SaaS provides complete software applications
    • D.IaaS provides virtual server infrastructure, PaaS provides complete software applications, and SaaS provides managed runtime platforms
    Show answerHide answer

    Correct answer: IaaS provides virtual server infrastructure, PaaS provides managed runtime platforms, and SaaS provides complete software applications

    The correct mapping is 'IaaS provides virtual server infrastructure, PaaS provides managed runtime platforms, and SaaS provides complete software applications'; moving from IaaS to PaaS to SaaS, the provider manages more of the stack and the customer controls less. 'IaaS provides complete software applications, PaaS provides virtual server infrastructure, and SaaS provides managed runtime platforms' and 'IaaS provides managed runtime platforms, PaaS provides complete software applications, and SaaS provides virtual server infrastructure' scramble all three models. 'IaaS provides virtual server infrastructure, PaaS provides complete software applications, and SaaS provides managed runtime platforms' gets IaaS right but swaps PaaS and SaaS, which are the two models candidates most often confuse.

  28. A company needs to store large volumes of unstructured data such as media files and backups, accessed over HTTP using unique keys and rich metadata rather than a traditional file hierarchy. Which storage type is most appropriate?

    • A.Block storage
    • B.File storage
    • C.Object storage
    • D.Ephemeral storage
    Show answerHide answer

    Correct answer: Object storage

    Object storage keeps each item as an object with a unique key and rich metadata in a flat namespace reached over HTTP APIs, which suits large volumes of unstructured data such as media files and backups. 'Block storage' exposes raw volumes to a single server for low-latency workloads like databases. 'File storage' presents a shared folder hierarchy over NFS or SMB, the traditional file hierarchy the stem rules out. 'Ephemeral storage' is temporary instance disk that is lost when the instance stops, so it cannot hold backups.

  29. A database server in the cloud requires a high-performance volume that the operating system can format with a file system and treats like a directly attached disk, supporting frequent random reads and writes. Which storage type meets this requirement?

    • A.File storage
    • B.Object storage
    • C.Block storage
    • D.Archive storage
    Show answerHide answer

    Correct answer: Block storage

    Block storage presents raw, fixed-size volumes that the operating system formats with its own file system and treats like a locally attached disk, giving the low latency and random read/write performance a database needs. 'File storage' adds a shared network file system layer on top, so the server mounts a share rather than formatting a disk. 'Object storage' is reached through HTTP APIs as whole objects and cannot be formatted or updated block by block. 'Archive storage' is a low-cost tier with slow retrieval, the opposite of high performance.

  30. Several virtual machines need to mount the same shared directory simultaneously using protocols such as NFS or SMB so multiple servers can read and write common files. Which cloud storage type provides this?

    • A.Block storage
    • B.File storage
    • C.Object storage
    • D.Direct-attached storage
    Show answerHide answer

    Correct answer: File storage

    File storage provides a shared, hierarchical file system that many clients can mount at the same time over NFS or SMB, so several servers can read and write the same files. 'Block storage' volumes are normally attached to one instance at a time and carry that instance's own file system. 'Object storage' is accessed through HTTP APIs rather than mounted as a shared directory. 'Direct-attached storage' is cabled to a single server and is not shared over the network at all.

  31. A cloud architect compares three storage paradigms. Which mapping of storage type to its primary characteristic is correct?

    • A.Block storage provides mountable folder hierarchies; object storage provides objects addressed by unique keys with metadata; file storage provides raw disk volumes
    • B.Block storage provides objects addressed by unique keys with metadata; object storage provides mountable folder hierarchies; file storage provides raw disk volumes
    • C.Block storage provides raw disk volumes; object storage provides mountable folder hierarchies; file storage provides objects addressed by unique keys with metadata
    • D.Block storage provides raw disk volumes; object storage provides objects addressed by unique keys with metadata; file storage provides mountable folder hierarchies
    Show answerHide answer

    Correct answer: Block storage provides raw disk volumes; object storage provides objects addressed by unique keys with metadata; file storage provides mountable folder hierarchies

    The correct mapping is 'Block storage provides raw disk volumes; object storage provides objects addressed by unique keys with metadata; file storage provides mountable folder hierarchies': block storage is disk-level and low-latency, object storage is API-based and scales for unstructured data, and file storage is shared and hierarchical. 'Block storage provides mountable folder hierarchies; object storage provides objects addressed by unique keys with metadata; file storage provides raw disk volumes' gets object storage right but swaps block and file storage. 'Block storage provides objects addressed by unique keys with metadata; object storage provides mountable folder hierarchies; file storage provides raw disk volumes' mismatches all three types. 'Block storage provides raw disk volumes; object storage provides mountable folder hierarchies; file storage provides objects addressed by unique keys with metadata' gets block storage right but swaps object and file storage.

  32. A storage policy automatically moves data that has not been accessed in 90 days from high-performance SSD-backed storage to a lower-cost archival tier. What is this practice called?

    • A.Storage replication
    • B.Storage tiering
    • C.Thin provisioning
    • D.Thick provisioning
    Show answerHide answer

    Correct answer: Storage tiering

    Storage tiering places data on different classes of storage according to how often it is used, automatically moving cold data from fast SSD-backed storage to cheaper archival tiers while hot data stays on fast media. 'Storage replication' copies data to a second location for resilience but does not move it between performance tiers. 'Thin provisioning' allocates physical capacity only as data is written, and 'Thick provisioning' reserves all capacity up front; both decide how space is allocated, not where aging data lives.

  33. A storage volume is advertised as delivering 16,000 IOPS. In the context of cloud storage performance, what does IOPS measure?

    • A.The count of read and write operations the volume can complete each second
    • B.The number of megabytes sent between a volume and its instance each second
    • C.The length of the delay between sending a request and getting its response
    • D.The ratio of hits and misses among requests sent to the cache
    Show answerHide answer

    Correct answer: The count of read and write operations the volume can complete each second

    IOPS stands for input/output operations per second: 'The count of read and write operations the volume can complete each second', a key measure for transactional workloads such as databases. 'The number of megabytes sent between a volume and its instance each second' is throughput, measured in MB/s rather than in operations. 'The length of the delay between sending a request and getting its response' is latency. 'The ratio of hits and misses among requests sent to the cache' is the cache hit ratio, which can raise IOPS but is not what IOPS counts.

  34. A cloud engineer is asked to define a virtual machine for a new deployment. Which description best captures what a virtual machine is?

    • A.A software emulation of a physical computer that runs its own operating system on shared hardware
    • B.A lightweight image that shares the host kernel and holds a single application with its libraries
    • C.A thin control layer that divides the hardware on one host among multiple guest operating systems
    • D.A saved copy of a disk image that serves as a template for launching new server instances
    Show answerHide answer

    Correct answer: A software emulation of a physical computer that runs its own operating system on shared hardware

    A virtual machine is 'A software emulation of a physical computer that runs its own operating system on shared hardware', with a hypervisor allocating the underlying CPU, memory and devices. 'A lightweight image that shares the host kernel and holds a single application with its libraries' describes a container. 'A thin control layer that divides the hardware on one host among multiple guest operating systems' describes the hypervisor that runs virtual machines, not the machine itself. 'A saved copy of a disk image that serves as a template for launching new server instances' describes a VM image or template, which is used to create virtual machines but is not a running machine.

  35. A type 1 hypervisor and a type 2 hypervisor are being compared for a production virtualization platform. What is the defining difference between them?

    • A.A type 1 hypervisor runs as an application on a host operating system, while a type 2 runs directly on the physical hardware
    • B.A type 1 hypervisor runs one virtual machine at a time, while a type 2 can host many isolated guest systems
    • C.A type 1 hypervisor runs directly on the physical hardware, while a type 2 runs as an application on a host operating system
    • D.A type 1 hypervisor can host many isolated guest systems, while a type 2 runs one virtual machine at a time
    Show answerHide answer

    Correct answer: A type 1 hypervisor runs directly on the physical hardware, while a type 2 runs as an application on a host operating system

    The defining difference is placement in the stack: 'A type 1 hypervisor runs directly on the physical hardware, while a type 2 runs as an application on a host operating system'. That bare-metal placement is why type 1 hypervisors are preferred for production platforms. 'A type 1 hypervisor runs as an application on a host operating system, while a type 2 runs directly on the physical hardware' reverses the two types. 'A type 1 hypervisor runs one virtual machine at a time, while a type 2 can host many isolated guest systems' and 'A type 1 hypervisor can host many isolated guest systems, while a type 2 runs one virtual machine at a time' are both wrong because both types run many virtual machines at once; the number of guests is not what separates them.

  36. In a virtualized cloud host, which component is responsible for abstracting the physical CPU, memory, and devices so that multiple isolated virtual machines can share the same hardware?

    • A.The container runtime
    • B.The hypervisor
    • C.The firmware
    • D.The guest operating system
    Show answerHide answer

    Correct answer: The hypervisor

    The hypervisor is the virtualization layer that abstracts the physical CPU, memory and devices and shares them out so several isolated virtual machines can run on one host. 'The container runtime' isolates processes that share a single host kernel; it does not present virtual hardware to whole machines. 'The firmware' (BIOS or UEFI) initializes the physical hardware at boot but does not divide it among virtual machines. 'The guest operating system' runs inside each virtual machine on top of the virtual hardware that the hypervisor provides.

  37. An application is being redesigned so that each business capability is built as a small, independently deployable service that communicates with the others over lightweight APIs. What architectural style is this?

    • A.A microservices architecture
    • B.A monolithic architecture
    • C.An N-tier architecture
    • D.An event-driven architecture
    Show answerHide answer

    Correct answer: A microservices architecture

    'A microservices architecture' splits an application into small services, each owning one business capability, that are deployed independently and talk to each other through lightweight APIs. 'A monolithic architecture' builds and deploys the whole application as one unit, even when its code is organized into modules. 'An N-tier architecture' separates presentation, logic and data layers, and each tier can still be one large deployable. 'An event-driven architecture' describes components reacting to published events; it is a communication style, not the decomposition into independently deployable services that the stem describes.

  38. A team builds an application specifically to exploit cloud capabilities, using containers, microservices, declarative APIs, and automated scaling rather than lifting an existing on-premises app unchanged. This approach is best described as what?

    • A.Cloud native
    • B.Cloud bursting
    • C.Lift and shift
    • D.Lift and reshape
    Show answerHide answer

    Correct answer: Cloud native

    Cloud native describes applications designed from the start to exploit cloud elasticity, typically with containers, microservices, declarative APIs and automated scaling. 'Cloud bursting' sends overflow demand from a private environment to a public cloud and says nothing about how the application is built. 'Lift and shift' moves an existing application to the cloud unchanged, the opposite of what the stem describes. 'Lift and reshape' (replatforming) moves an existing application with a few optimizations, such as a managed database, rather than building it for the cloud.

  39. An application that processes uploaded images runs only when a file is uploaded, scales automatically per request, and incurs cost only while code executes, with no servers for the team to provision or patch. Which compute model is this?

    • A.Edge computing
    • B.Instance autoscaling
    • C.Serverless computing
    • D.Spot instance pools
    Show answerHide answer

    Correct answer: Serverless computing

    Serverless computing runs code only when an event such as a file upload triggers it, scales automatically per request and bills only for execution time, while the provider manages the servers. 'Edge computing' moves processing closer to where data is produced to cut latency but still needs compute to be provisioned and managed. 'Instance autoscaling' adds and removes virtual machines with load, yet the team still patches those instances and pays while they run. 'Spot instance pools' use discounted spare capacity that can be reclaimed, but they are still virtual machines the team manages.

  40. A logically isolated, software-defined network segment within a public cloud provider, where a customer controls its own private IP address range, subnets, and routing, is known as what?

    • A.A private cloud
    • B.A virtual private cloud (VPC)
    • C.A perimeter network (DMZ)
    • D.A software-defined perimeter (SDP)
    Show answerHide answer

    Correct answer: A virtual private cloud (VPC)

    A virtual private cloud (VPC) is a logically isolated, software-defined network within a public cloud in which the customer chooses its own private IP address range, subnets, route tables and gateways. 'A private cloud' is a deployment model in which cloud infrastructure is dedicated to one organization, often on its own premises, rather than an isolated network carved out inside a public provider. 'A perimeter network (DMZ)' is a screened segment that exposes public-facing services, usually built as one subnet inside a VPC rather than being the VPC itself. 'A software-defined perimeter (SDP)' is a Zero Trust access model that hides applications until each user and device is authenticated; it controls access but does not give the customer its own address range, subnets and routing.

  41. Within a virtual network, an engineer divides the address space into smaller logical segments to organize resources and control traffic flow. What is each of these smaller segments called?

    • A.A router
    • B.A subnet
    • C.A security group
    • D.A peering connection
    Show answerHide answer

    Correct answer: A subnet

    A subnet is a smaller logical segment carved out of a network's address range, used to group resources and control how traffic flows within a virtual network. 'A router' forwards traffic between network segments rather than being one of the segments. 'A security group' is a virtual firewall attached to instances rather than a slice of the address range. 'A peering connection' links two separate virtual networks together instead of dividing one.

  42. A network is defined using the address 10.0.1.0/24. In CIDR notation, what does the /24 indicate, and how many usable host addresses does this provide?

    • A.The first 24 bits identify the network, leaving 254 usable addresses
    • B.The first 24 bits identify the network, leaving 256 usable addresses
    • C.The last 24 bits identify the host, leaving 16,777,214 usable addresses
    • D.The last 24 bits identify the host, leaving 16,777,216 usable addresses
    Show answerHide answer

    Correct answer: The first 24 bits identify the network, leaving 254 usable addresses

    In CIDR notation the /24 is the prefix length: 'The first 24 bits identify the network, leaving 254 usable addresses', because the remaining 8 host bits give 256 addresses, minus the network and broadcast addresses. 'The first 24 bits identify the network, leaving 256 usable addresses' forgets to subtract those two reserved addresses. 'The last 24 bits identify the host, leaving 16,777,214 usable addresses' reads the prefix as host bits (2242^{24} minus 2), and 'The last 24 bits identify the host, leaving 16,777,216 usable addresses' makes that same mistake and also skips the reserved addresses.

  43. In a virtual private cloud, an engineer must place internet-facing web servers and internal-only database servers appropriately. What primarily distinguishes a public subnet from a private subnet?

    • A.A public subnet uses NAT for its outbound traffic, while a private subnet routes out to an internet gateway
    • B.A public subnet is shared with other cloud customers by default, while a private subnet has a single tenant
    • C.A public subnet accepts all inbound traffic by default, while a private subnet blocks it
    • D.A public subnet routes to an internet gateway for internet access, while a private subnet has no such route
    Show answerHide answer

    Correct answer: A public subnet routes to an internet gateway for internet access, while a private subnet has no such route

    The defining difference is routing: 'A public subnet routes to an internet gateway for internet access, while a private subnet has no such route', which is why web servers go in public subnets and databases in private ones. 'A public subnet uses NAT for its outbound traffic, while a private subnet routes out to an internet gateway' reverses the design; the NAT gateway sits in the public subnet so private instances can reach out. 'A public subnet is shared with other cloud customers by default, while a private subnet has a single tenant' is wrong because every subnet in a VPC belongs to that customer's account. 'A public subnet accepts all inbound traffic by default, while a private subnet blocks it' is wrong because security groups and network ACLs filter traffic in both kinds of subnet.

  44. A media company wants users worldwide to download large video files with low latency by serving them from edge locations geographically close to each user rather than always from the origin server. Which technology provides this?

    • A.A virtual private network (VPN)
    • B.A content delivery network (CDN)
    • C.A global server load balancer (GSLB)
    • D.An application load balancer (ALB)
    Show answerHide answer

    Correct answer: A content delivery network (CDN)

    A content delivery network (CDN) caches content at edge locations around the world and serves each user from a nearby edge, cutting latency and offloading the origin server. 'A global server load balancer (GSLB)' steers users to the nearest healthy data center or region, but it does not cache the files at edge locations, so downloads still come from full servers. 'An application load balancer (ALB)' spreads requests across servers within one region. 'A virtual private network (VPN)' encrypts traffic over an untrusted network and adds overhead rather than bringing content closer to users.

  45. A network team wants to centrally manage and program network behavior in software, decoupling the control plane that makes forwarding decisions from the data plane that moves packets. What is this approach called?

    • A.Network function virtualization (NFV)
    • B.Network address translation (NAT)
    • C.Policy-based routing (PBR)
    • D.Software-defined networking (SDN)
    Show answerHide answer

    Correct answer: Software-defined networking (SDN)

    Software-defined networking (SDN) separates the control plane, which decides how traffic is forwarded, from the data plane, which moves the packets, so the network can be managed and programmed centrally through software. 'Network function virtualization (NFV)' runs network functions such as firewalls and load balancers as software on standard servers, but it does not by itself split the control plane from the data plane. 'Network address translation (NAT)' rewrites IP addresses as traffic crosses a boundary. 'Policy-based routing (PBR)' forwards selected traffic by rules such as source address instead of the routing table, but each router still holds and applies its own rules.

  46. A small architecture team must explain the difference between scalability and elasticity. Which statement is correct?

    • A.Scalability is the ability to grow by adding more power to one existing machine, while elasticity is the ability to grow by adding servers to a shared pool
    • B.Scalability is the ability to handle growth by adding capacity, while elasticity is the ability to add and remove resources automatically as demand changes
    • C.Elasticity is the ability to handle growth by adding capacity, while scalability is the ability to add and remove resources automatically as demand changes
    • D.Elasticity is the ability to grow by adding more power to one existing machine, while scalability is the ability to grow by adding servers to a shared pool
    Show answerHide answer

    Correct answer: Scalability is the ability to handle growth by adding capacity, while elasticity is the ability to add and remove resources automatically as demand changes

    'Scalability is the ability to handle growth by adding capacity, while elasticity is the ability to add and remove resources automatically as demand changes': scalability is the capacity to grow, and elasticity is automatic adjustment in both directions, close to real time. 'Elasticity is the ability to handle growth by adding capacity, while scalability is the ability to add and remove resources automatically as demand changes' swaps the two definitions. 'Scalability is the ability to grow by adding more power to one existing machine, while elasticity is the ability to grow by adding servers to a shared pool' and 'Elasticity is the ability to grow by adding more power to one existing machine, while scalability is the ability to grow by adding servers to a shared pool' confuse the two terms with vertical and horizontal scaling; both terms can involve either direction, and only elasticity implies automatic scaling back down.

Deployment (41)

  1. In cloud deployment, what is the primary function of a cloud gateway?

    • A.To route traffic between separate cloud environments
    • B.To spread requests between redundant cloud instances
    • C.To filter packets between isolated cloud subnets
    • D.To replicate snapshots between paired cloud regions
    Show answerHide answer

    Correct answer: To route traffic between separate cloud environments

    To route traffic between separate cloud environments is the function of a cloud gateway: it gives on-premises systems and different cloud platforms a single point of connectivity for integration and data migration. To spread requests between redundant cloud instances is the job of a load balancer. To filter packets between isolated cloud subnets describes a firewall or security group. To replicate snapshots between paired cloud regions is a data-protection task handled by backup and replication services.

  2. In cloud management, what is the main purpose of using 'Infrastructure as Code' IaC?

    • A.To encrypt and protect data moving through the infrastructure
    • B.To provision and manage the infrastructure automatically
    • C.To monitor and analyze traffic crossing the infrastructure
    • D.To grant and revoke the access rights of infrastructure users
    Show answerHide answer

    Correct answer: To provision and manage the infrastructure automatically

    Infrastructure as Code is used to provision and manage the infrastructure automatically from version-controlled definition files, so environments can be built the same way every time. 'To encrypt and protect data moving through the infrastructure' is the job of TLS and other encryption controls; 'to monitor and analyze traffic crossing the infrastructure' is network monitoring; 'to grant and revoke the access rights of infrastructure users' is identity and access management.

  3. An organization is moving an on-premises application to the cloud with minimal changes, essentially performing a 'lift and shift.' Which migration strategy does this describe?

    • A.Retain
    • B.Retire
    • C.Rehost
    • D.Replatform
    Show answerHide answer

    Correct answer: Rehost

    Rehost is the 'lift and shift' strategy: the application moves to cloud infrastructure with little or no change to its code or architecture. Retain means leaving the application where it is for now, so nothing moves. Retire means decommissioning an application that is no longer needed, so it is not migrated at all. Replatform makes a few cloud optimizations during the move, such as switching to a managed database, which is more than this scenario describes.

  4. A team wants to migrate an application to the cloud while making minor optimizations, such as switching to a managed database service, without changing the core application code. Which strategy is this?

    • A.Replatform
    • B.Rehost
    • C.Retain
    • D.Re-architect
    Show answerHide answer

    Correct answer: Replatform

    Replatform (sometimes called 'lift, tinker and shift') is the strategy described: the core code stays the same while a few components are swapped for cloud services, such as a managed database. Rehost moves the application without any of those optimizations. Retain leaves the application where it is, so it is not migrated yet. Re-architect (also called refactor) redesigns the application and changes its code to use cloud-native features, which this team wants to avoid.

  5. During migration planning, an organization decides that a legacy application will not move to the cloud and will be decommissioned because it is no longer needed. Which migration disposition is this?

    • A.Retire
    • B.Retain
    • C.Rehost
    • D.Repurchase
    Show answerHide answer

    Correct answer: Retire

    Retire is the disposition for an application that is no longer needed: it is decommissioned instead of migrated, which shrinks the migration's scope and its running costs. Retain keeps an application running where it is, usually on premises, because it is still needed but not ready to move. Rehost moves the application to the cloud unchanged, so it keeps running. Repurchase swaps the application for a new product, often SaaS, so the business still has the capability.

  6. What is the primary characteristic of Infrastructure as Code (IaC)?

    • A.Configuring servers manually from web consoles
    • B.Executing event-driven code from serverless functions
    • C.Provisioning resources from machine-readable definition files
    • D.Restoring environments from scheduled point-in-time snapshots
    Show answerHide answer

    Correct answer: Provisioning resources from machine-readable definition files

    Provisioning resources from machine-readable definition files is the defining trait of IaC: templates written in formats such as JSON, YAML or HCL describe the desired infrastructure, so it can be versioned, reviewed and deployed the same way every time. Configuring servers manually from web consoles is the hand-built approach that IaC replaces. Executing event-driven code from serverless functions describes serverless computing (FaaS), not how infrastructure is defined. Restoring environments from scheduled point-in-time snapshots is disaster recovery, which brings back saved data rather than building infrastructure from code.

  7. An IaC deployment is failing because the actual state of the cloud environment no longer matches the state declared in the configuration files. What is this condition called?

    • A.Configuration management
    • B.Configuration drift
    • C.State file locking
    • D.State file corruption
    Show answerHide answer

    Correct answer: Configuration drift

    Configuration drift is the name for this condition: changes made outside the IaC workflow, such as manual edits in the console, leave the real environment different from what the code declares, and drift detection is used to find and reconcile them. Configuration management is the discipline (and the tools, such as Ansible or Puppet) used to keep systems in their desired state; it is the remedy, not the condition. State file locking stops two runs from changing the same state at once, so a locked state blocks a deployment for a different reason. State file corruption means the tool's own record of what it manages is damaged or unreadable, which is a different failure from the live environment diverging from the code.

  8. Which two data serialization formats are most commonly used to author Infrastructure as Code templates?

    • A.XML and CSV
    • B.INI and TOML
    • C.JSON and YAML
    • D.HTML and DOCX
    Show answerHide answer

    Correct answer: JSON and YAML

    JSON and YAML are the formats most IaC templates are written in: AWS CloudFormation accepts both, Azure ARM templates use JSON, Kubernetes manifests and Ansible playbooks use YAML, and Terraform also reads a JSON variant of its configuration. XML and CSV are general data formats; XML has largely fallen out of use for infrastructure templates, and CSV holds flat tables that cannot express nested resources. INI and TOML are simple configuration-file formats for applications, not the usual template formats for cloud provisioning. HTML and DOCX are formats for documents and web pages, not for describing infrastructure.

  9. A cloud engineer is preconfiguring a virtual machine image with a hardened OS, patches, and required software so that all new instances launch from a consistent, approved baseline. What is this image commonly called?

    • A.Container image
    • B.Golden image
    • C.Volume snapshot
    • D.Boot volume
    Show answerHide answer

    Correct answer: Golden image

    Golden image is the term for a hardened, fully patched and preconfigured base image from which every new instance is launched, so all instances start from the same approved state. A container image packages an application and its dependencies to run as a container, not as a full virtual machine with its own OS. A volume snapshot is a point-in-time copy of a disk taken for backup or recovery; it captures whatever state the disk was in, not a curated baseline. A boot volume is the disk an individual instance starts from, unique to that instance rather than a shared standard.

  10. When provisioning storage for a high-transaction database, which performance metric is the most important to evaluate?

    • A.MIPS (millions of instructions per second)
    • B.IOPS (input/output operations per second)
    • C.Gb/s (gigabits per second)
    • D.Mpps (million packets per second)
    Show answerHide answer

    Correct answer: IOPS (input/output operations per second)

    IOPS (input/output operations per second) is the metric that matters most, because a high-transaction database performs many small, random reads and writes, and the volume must sustain enough of them (for example, a provisioned-IOPS SSD volume). MIPS (millions of instructions per second) measures how fast a processor executes instructions, a compute property, not a storage one. Gb/s (gigabits per second) measures how much data a network link carries, not how many storage operations the volume can handle. Mpps (million packets per second) is a network-device metric for routers, firewalls and load balancers, not a measure of storage performance.

  11. An engineer needs to provision a network boundary in the cloud that isolates a set of resources with its own IP address range, subnets, and routing. What should be deployed?

    • A.A virtual private network (VPN)
    • B.A virtual private server (VPS)
    • C.A virtual private gateway (VGW)
    • D.A virtual private cloud (VPC)
    Show answerHide answer

    Correct answer: A virtual private cloud (VPC)

    A virtual private cloud (VPC) is what should be deployed: it is a logically isolated network in the provider's cloud with its own IP address range, subnets, route tables and gateways. A virtual private network (VPN) is an encrypted tunnel that connects sites or users to a network; it does not create an isolated network with its own subnets. A virtual private server (VPS) is a single virtual machine rented from a hosting provider, a compute resource rather than a network boundary. A virtual private gateway (VGW) is the VPN endpoint attached to a VPC; it connects the VPC to another network but is not the isolated network itself.

  12. A company wants to deploy resources across multiple isolated data center locations within a single cloud region to improve fault tolerance. Which construct should they use?

    • A.Several network security groups
    • B.Several availability zones
    • C.Several edge locations
    • D.Several cross-region replicas
    Show answerHide answer

    Correct answer: Several availability zones

    Several availability zones are the right construct: availability zones are physically separate data centers with their own power and networking inside one region, so a failure in one zone leaves the others running. Several network security groups only add more firewall rule sets and do nothing for physical fault tolerance. Several edge locations belong to a content delivery network and cache content near users; they are not places to run the workload inside the region. Several cross-region replicas copy data to other regions, which goes beyond the single-region design the company described.

  13. What does 'right-sizing' refer to when provisioning cloud compute resources?

    • A.Matching server capacity to actual workload needs
    • B.Committing predictable load to reserved instances
    • C.Moving idle data volumes to cheaper storage tiers
    • D.Shifting incoming requests to healthy targets
    Show answerHide answer

    Correct answer: Matching server capacity to actual workload needs

    Matching server capacity to actual workload needs is what right-sizing means: choosing the instance type, CPU and memory that the measured workload requires, avoiding both wasted over-provisioning and slow under-provisioning. Committing predictable load to reserved instances lowers the price per hour, but it does not change whether each instance is the right size. Moving idle data volumes to cheaper storage tiers is storage tiering, which concerns storage, not compute capacity. Shifting incoming requests to healthy targets is the job of a load balancer.

  14. Which deployment approach reduces risk by gradually shifting a small percentage of production traffic to a new version before a full rollout?

    • A.Rolling deployment
    • B.Blue-green deployment
    • C.Canary deployment
    • D.Big bang deployment
    Show answerHide answer

    Correct answer: Canary deployment

    Canary deployment is the approach described: a small share of real traffic goes to the new version first, and the share grows only while monitoring shows no problems. Rolling deployment replaces instances in batches until all run the new version, so progress is measured in servers updated rather than in a controlled slice of traffic. Blue-green deployment switches all traffic from one complete environment to another in a single step. Big bang deployment releases the new version to everyone at once, which carries the most risk.

  15. In a blue-green deployment, what is the primary advantage?

    • A.It limits deployment risk by exposing changes to the smallest user group
    • B.It saves hardware costs by upgrading existing servers to the latest code
    • C.It tests releases by mirroring live requests to the candidate version
    • D.It gives near-instant rollback by routing traffic to the old environment
    Show answerHide answer

    Correct answer: It gives near-instant rollback by routing traffic to the old environment

    'It gives near-instant rollback by routing traffic to the old environment' is the main advantage: blue-green keeps two identical environments, sends production traffic to the new one after validation, and keeps the previous one ready, so a single switch undoes a bad release. 'It limits deployment risk by exposing changes to the smallest user group' describes a canary release, not blue-green, which moves all traffic at once. 'It saves hardware costs by upgrading existing servers to the latest code' describes a rolling deployment; blue-green actually needs a second full environment. 'It tests releases by mirroring live requests to the candidate version' describes a shadow deployment, where the copied traffic never reaches real users.

  16. An engineer is using a deployment template that accepts input variables so the same template can provision dev, test, and production environments. What IaC benefit does this demonstrate?

    • A.Parameterization and repeatability
    • B.Auditability and version control
    • C.Drift detection and automatic remediation
    • D.Compliance and governance
    Show answerHide answer

    Correct answer: Parameterization and repeatability

    Parameterization and repeatability are what the scenario shows: input variables let one template build dev, test and production in the same way, with only the values changing. Auditability and version control come from keeping templates in a repository with a change history, not from input variables. Drift detection and automatic remediation compare the live environment with the code and correct differences, which is a separate practice. Compliance and governance come from checking templates against rules (policy as code) and controlling who may change them, not from reusing one template with different inputs.

  17. A cloud team needs to deploy short-lived, event-driven code without provisioning or managing any servers, paying only for execution time. Which resource type should they provision?

    • A.A spot instance
    • B.A reserved instance
    • C.A dedicated bare-metal host
    • D.A serverless function
    Show answerHide answer

    Correct answer: A serverless function

    A serverless function fits every requirement: the provider runs the code in response to events, scales it automatically and bills only for invocations and execution time, with no servers to provision or manage. A spot instance is cheap spare capacity, but it is still a virtual machine that must be configured and managed, and it can be reclaimed at short notice. A reserved instance is a VM bought on a one- or three-year commitment, billed whether or not code runs. A dedicated bare-metal host is a whole physical server that the customer must manage, which is the opposite of what the team wants.

  18. Before promoting an IaC configuration to production, a team runs automated validation to confirm the template syntax and planned changes are correct. Which IaC practice does this represent?

    • A.Tagging
    • B.Logging
    • C.Testing
    • D.State locking
    Show answerHide answer

    Correct answer: Testing

    Testing is the practice described: automated checks such as syntax validation, linting and a plan or preview step confirm that the template is valid and that its changes match what was intended before anything reaches production. Tagging adds metadata to resources for cost tracking and organization. Logging records what happened during and after a deployment, which helps investigate problems but does not validate a template beforehand. State locking stops two people or pipelines from changing the same IaC state at the same time.

  19. Which cloud networking component allows instances in a private subnet to initiate outbound internet connections while preventing inbound connections from the internet?

    • A.Internet gateway
    • B.VPN gateway
    • C.NAT gateway
    • D.Customer gateway
    Show answerHide answer

    Correct answer: NAT gateway

    NAT gateway is correct: it translates private addresses so instances in a private subnet can start outbound connections, for example to download updates, while unsolicited inbound connections from the internet are dropped. An internet gateway gives a public subnet two-way internet access, so on its own it would allow inbound connections. A VPN gateway terminates encrypted site-to-site tunnels to another private network rather than giving instances general internet access. A customer gateway is the on-premises end of a site-to-site VPN connection, not a component inside the cloud network.

  20. When provisioning resources, an engineer applies metadata key-value pairs to track cost center, environment, and owner. What is this practice called?

    • A.Chargeback
    • B.Resource locking
    • C.Resource grouping
    • D.Tagging
    Show answerHide answer

    Correct answer: Tagging

    Tagging is the practice of attaching key-value metadata, such as cost center, environment and owner, to cloud resources, so they can be filtered, reported on and automated. Chargeback bills each department for the cloud costs it uses; it depends on tags but is a financial process, not the act of labeling resources. Resource locking (for example, an Azure resource lock) stops resources from being deleted or changed by mistake; it protects them but records nothing about cost center or owner. Resource grouping collects related resources into a container (such as an Azure resource group) to manage them together, rather than adding key-value pairs to each one.

  21. A cloud architect must select a deployment that requires migrating a monolithic application into independently deployable services to fully leverage cloud-native scaling. Which migration strategy is this?

    • A.Replatform
    • B.Refactor
    • C.Rehost
    • D.Retire
    Show answerHide answer

    Correct answer: Refactor

    'Refactor' (also called re-architect) is the strategy that redesigns an application, typically splitting a monolith into independently deployable microservices, so it can use cloud-native capabilities such as elastic scaling; it delivers the most long-term benefit for the most effort. 'Replatform' makes a few targeted optimizations, such as moving to a managed database, while leaving the core architecture intact. 'Rehost' moves the application as-is (lift and shift) with no redesign. 'Retire' decommissions an application that is no longer needed, so nothing is migrated at all.

  22. A cloud engineer is provisioning a 500 GB block volume for a development virtual machine that will initially hold only about 40 GB of data. To minimize the physical capacity consumed up front while still presenting the full 500 GB to the guest OS, which storage provisioning method should be selected?

    • A.Thick provisioning
    • B.Data deduplication
    • C.Thin provisioning
    • D.Data compression
    Show answerHide answer

    Correct answer: Thin provisioning

    Thin provisioning presents the full 500 GB to the guest but consumes physical capacity only as data is actually written, so a volume holding about 40 GB uses roughly 40 GB on the back end. 'Thick provisioning' reserves the whole 500 GB at creation, whether eager- or lazy-zeroed, which wastes capacity on a lightly used development volume. 'Data deduplication' removes duplicate blocks and 'Data compression' shrinks the data that is stored; both reduce what written data occupies, but neither is a provisioning method that decides how logical size maps to physical allocation.

  23. A storage administrator is comparing thick and thin provisioning for a new high-transaction production database. Which statement accurately describes the trade-off between the two approaches?

    • A.Thick provisioning allocates space on first write to improve storage density, while thin provisioning pre-allocates all blocks at creation to give predictable performance
    • B.Thick provisioning allocates space on first write to give predictable performance, while thin provisioning pre-allocates all blocks at creation to improve storage density
    • C.Thick provisioning pre-allocates all blocks at creation to improve storage density, while thin provisioning allocates space on first write to give predictable performance
    • D.Thick provisioning pre-allocates all blocks at creation to give predictable performance, while thin provisioning allocates space on first write to improve storage density
    Show answerHide answer

    Correct answer: Thick provisioning pre-allocates all blocks at creation to give predictable performance, while thin provisioning allocates space on first write to improve storage density

    'Thick provisioning pre-allocates all blocks at creation to give predictable performance, while thin provisioning allocates space on first write to improve storage density', which is why thick provisioning is usually preferred for high-I/O databases. 'Thick provisioning allocates space on first write to improve storage density, while thin provisioning pre-allocates all blocks at creation to give predictable performance' swaps the two methods completely. 'Thick provisioning allocates space on first write to give predictable performance, while thin provisioning pre-allocates all blocks at creation to improve storage density' swaps how each one allocates space. 'Thick provisioning pre-allocates all blocks at creation to improve storage density, while thin provisioning allocates space on first write to give predictable performance' gets the allocation right but swaps the benefits; pre-allocating everything costs density, and allocating on first write adds a small delay to new writes.

  24. A team using thin provisioning across a shared storage pool discovers that the total logical capacity promised to all volumes exceeds the physical capacity actually installed. What is this condition called, and what primary risk does it introduce?

    • A.Underprovisioning; each volume gets less capacity than its peak workload needs from the pool
    • B.Storage sprawl; idle volumes and stale snapshots still hold paid capacity in the shared pool
    • C.Write amplification; garbage collection wears out the flash drives in the pool early
    • D.Oversubscription; volume writes can fail if the shared pool fills up before new disks arrive
    Show answerHide answer

    Correct answer: Oversubscription; volume writes can fail if the shared pool fills up before new disks arrive

    Promising more logical capacity than is physically installed is oversubscription (also called overprovisioning), and its main risk is 'Oversubscription; volume writes can fail if the shared pool fills up before new disks arrive', which is why oversubscribed pools need close capacity monitoring. 'Underprovisioning; each volume gets less capacity than its peak workload needs from the pool' is the opposite condition, where too little is allocated rather than too much promised. 'Storage sprawl; idle volumes and stale snapshots still hold paid capacity in the shared pool' describes waste from forgotten resources, not promises that exceed the hardware. 'Write amplification; garbage collection wears out the flash drives in the pool early' is an SSD wear effect, unrelated to how capacity is promised.

  25. An organization is moving its applications, data, and workloads from its on-premises data center into a public cloud provider's infrastructure. At a high level, what is this overall process called?

    • A.Cloud migration
    • B.Cloud repatriation
    • C.Cloud bursting
    • D.Cloud replication
    Show answerHide answer

    Correct answer: Cloud migration

    Cloud migration is the overall process of moving applications, data and workloads from an on-premises data center, or another environment, into a cloud provider's infrastructure. 'Cloud repatriation' is the reverse move, bringing workloads back from the cloud to on-premises infrastructure. 'Cloud bursting' temporarily sends overflow demand from a private environment to a public cloud while the workload's home stays on premises. 'Cloud replication' keeps copies of data or systems synchronized in the cloud for resilience, which is one technique a migration might use rather than the overall move.

  26. During migration planning, a cloud engineer must choose a migration strategy for a legacy application. The business wants it moved to the cloud as quickly as possible with no rewriting of code and minimal modification. Which migration approach fits this requirement?

    • A.Replatform
    • B.Repurchase
    • C.Rehost
    • D.Refactor
    Show answerHide answer

    Correct answer: Rehost

    Rehost, better known as lift and shift, moves the application to the cloud essentially as it is, with no code rewriting and minimal change, which makes it the fastest strategy. 'Replatform' adds small optimizations on the way, such as switching to a managed database, which is more change than the business wants. 'Repurchase' replaces the application with a SaaS product instead of moving the existing one. 'Refactor' re-architects the application, for example into microservices, and takes the most time and effort.

  27. A development team wants every code release to roll out to its fleet of identical web servers gradually: a few servers at a time are taken out of rotation, updated to the new version, and returned to service until the entire fleet is updated, using only the existing production environment. Which deployment strategy is this?

    • A.Canary deployment
    • B.Rolling deployment
    • C.Big-bang deployment
    • D.Blue-green deployment
    Show answerHide answer

    Correct answer: Rolling deployment

    A rolling deployment fits: it upgrades a few servers at a time inside the one existing production environment, taking each batch out of rotation and returning it once it runs the new version, until the whole fleet is done. A canary deployment sends a small share of users to the new version and watches it, rather than cycling the whole fleet batch by batch; a big-bang deployment upgrades every server at once; and a blue-green deployment builds a second full environment and moves all traffic to it in one switch.

  28. An engineer maintains two identical production-capacity environments. The current live version runs in 'blue' while the new version is deployed and validated in 'green'; once validated, all production traffic is switched from blue to green at once. What is the primary benefit emphasized by this approach over a gradual rollout?

    • A.Gradual exposure, since a small share of real users meets the new version first
    • B.Near-instant rollback, since a single switch puts users back on the untouched old release
    • C.Lower compute cost, since the update runs on current servers without a second environment
    • D.Smaller blast radius, since a fault reaches just the servers updated in each batch
    Show answerHide answer

    Correct answer: Near-instant rollback, since a single switch puts users back on the untouched old release

    The benefit blue-green emphasizes is 'near-instant rollback, since a single switch puts users back on the untouched old release': blue stays fully intact after the cutover, so if green misbehaves the router simply points traffic back. 'Gradual exposure, since a small share of real users meets the new version first' describes a canary release, the gradual approach being contrasted. 'Lower compute cost, since the update runs on current servers without a second environment' is wrong because blue-green keeps two production-sized environments and roughly doubles capacity during a release. 'Smaller blast radius, since a fault reaches just the servers updated in each batch' belongs to a rolling rollout; blue-green moves every user at once.

  29. A team releasing a payment service routes 5 percent of live user traffic to the new version while 95 percent continues on the stable version, then watches error rates and latency before widening the rollout to 25 percent and eventually 100 percent. Which deployment strategy is described, and what is its defining characteristic?

    • A.Canary deployment; a small share of live traffic validates the new build before full rollout
    • B.Blue-green deployment; a standby environment runs the new build and then takes the traffic at once
    • C.Rolling deployment; a handful of servers at a time get the new build and rejoin the fleet
    • D.In-place deployment; the running servers are halted and restarted with the new build
    Show answerHide answer

    Correct answer: Canary deployment; a small share of live traffic validates the new build before full rollout

    'Canary deployment; a small share of live traffic validates the new build before full rollout' is right: sending 5 percent of users to the new release, watching error rates and latency, then widening to 25 and 100 percent is the canary pattern. 'Blue-green deployment; a standby environment runs the new build and then takes the traffic at once' is a true description of blue-green, but the team never moves everyone in one switch. 'Rolling deployment; a handful of servers at a time get the new build and rejoin the fleet' is true of rolling, which replaces servers in batches rather than choosing a share of users to measure. 'In-place deployment; the running servers are halted and restarted with the new build' describes an in-place upgrade, which splits no traffic at all.

  30. A platform team is deciding between a blue-green and a canary deployment for a high-risk release. Which statement correctly distinguishes the two strategies?

    • A.Canary switches all users at once between two full environments, while blue-green shifts a small share of traffic to the new version first
    • B.Blue-green overwrites the running servers in place, while canary mirrors live requests to a hidden test copy whose replies are thrown away
    • C.Blue-green switches all users at once between two full environments, while canary shifts a small share of traffic to the new version first
    • D.Canary overwrites the running servers in place, while blue-green mirrors live requests to a hidden test copy whose replies are thrown away
    Show answerHide answer

    Correct answer: Blue-green switches all users at once between two full environments, while canary shifts a small share of traffic to the new version first

    'Blue-green switches all users at once between two full environments, while canary shifts a small share of traffic to the new version first' is the correct distinction: blue-green cuts everyone over in one move, and canary widens exposure step by step while metrics are watched. 'Canary switches all users at once between two full environments, while blue-green shifts a small share of traffic to the new version first' reverses the two. 'Blue-green overwrites the running servers in place, while canary mirrors live requests to a hidden test copy whose replies are thrown away' and its mirror 'canary overwrites the running servers in place, while blue-green mirrors live requests to a hidden test copy whose replies are thrown away' describe an in-place upgrade and a shadow deployment, which are neither strategy.

  31. A cloud engineer needs to provision compute that runs only in response to events, scales automatically per request, and bills based on execution time with no servers to manage. The team will package the code and let the provider handle the runtime. Which compute provisioning model best fits?

    • A.Function as a Service (FaaS)
    • B.Infrastructure as a Service (IaaS)
    • C.Container as a Service (CaaS)
    • D.Platform as a Service (PaaS)
    Show answerHide answer

    Correct answer: Function as a Service (FaaS)

    Function as a Service (FaaS), or serverless functions, fits: code runs only when an event triggers it, scales per request, and is billed by execution time while the provider runs everything underneath. Infrastructure as a Service (IaaS) hands the team virtual machines to size, patch and pay for while they sit idle. Container as a Service (CaaS) still asks the team to package and schedule containers on a cluster. Platform as a Service (PaaS) also hides the runtime, but its application instances stay running and are billed whether or not events arrive.

  32. During a migration assessment, an engineer documents an application that will be neither moved nor modernized because the team plans to leave it running on-premises for now and revisit it later. Within the common migration strategy framework, which disposition does this represent?

    • A.Retire (shut it down)
    • B.Retain (keep it local)
    • C.Rehost (move it as is)
    • D.Refactor (redesign it)
    Show answerHide answer

    Correct answer: Retain (keep it local)

    Retain (keep it local) is the disposition for an application the team deliberately leaves where it is, neither migrated nor modernized, to look at again later. Retire (shut it down) decommissions an application that is no longer needed, but this one keeps running. Rehost (move it as is), the lift-and-shift strategy, moves the workload to the cloud unchanged, and refactor (redesign it) re-architects it for cloud-native services; both act on the application now.

  33. A cloud engineer is provisioning a managed load balancer to distribute incoming HTTPS requests across a group of application servers. To verify that the load balancer only routes traffic to instances capable of serving requests, what must the engineer configure?

    • A.Connection draining on the target group
    • B.Health checks on the target group
    • C.Session affinity on the target group
    • D.Least-connections routing on the target group
    Show answerHide answer

    Correct answer: Health checks on the target group

    Health checks on the target group are required: the load balancer probes each backend at a set interval and sends requests only to targets that answer as healthy, pulling failed ones out of rotation. Connection draining on the target group lets in-flight requests finish when a target is removed, but it does not decide which targets are fit to serve. Session affinity on the target group pins a client to one target, and least-connections routing on the target group picks the least busy target; both would still send traffic to an instance that has stopped responding.

  34. An engineer must provision object storage for a static website's images and downloadable PDFs, where files are written once and read frequently over the public web by their unique keys. Which storage type is the most appropriate choice for this provisioning task?

    • A.File storage
    • B.Block storage
    • C.Object storage
    • D.Ephemeral storage
    Show answerHide answer

    Correct answer: Object storage

    Object storage suits write-once, read-often web content: each file is stored as an object under a unique key, served over HTTP, scales without limit and pairs naturally with a CDN. Block storage is a raw disk volume attached to one virtual machine and cannot serve files to the web on its own. File storage is a shared network file system for servers that need POSIX-style access, not public web delivery. Ephemeral storage is temporary instance disk that is lost when the instance stops.

  35. While integrating a newly provisioned application into the cloud environment, an engineer needs the app to retrieve a database password securely at runtime without hardcoding the credential into the deployment template or image. Which approach best meets this requirement during deployment?

    • A.Write the database password inside the template as a hidden parameter default
    • B.Type the database password into the template as a plaintext environment variable
    • C.Retrieve the database password from a secrets management service at runtime
    • D.Store the database password in a protected file inside the golden image
    Show answerHide answer

    Correct answer: Retrieve the database password from a secrets management service at runtime

    Retrieve the database password from a secrets management service at runtime is the secure approach: the application fetches the value from a managed secrets vault, such as AWS Secrets Manager, Azure Key Vault or HashiCorp Vault, only when it needs it, so the password never sits in the template, the image or the repository, and access can be audited and rotated. 'Write the database password inside the template as a hidden parameter default' only masks the value on screen; the default still lives in the template file. 'Type the database password into the template as a plaintext environment variable' also hardcodes it, because the template carries the literal value. 'Store the database password in a protected file inside the golden image' bakes the secret into every copy of the image.

  36. An engineer is provisioning network infrastructure and must allow virtual machines in a private subnet to download OS updates from the internet while preventing any inbound connections initiated from the internet to those VMs. Which component satisfies this requirement?

    • A.A NAT gateway placed in a public subnet
    • B.An internet gateway attached to a private subnet
    • C.A transit gateway placed in a public subnet
    • D.A VPN gateway attached to a private subnet
    Show answerHide answer

    Correct answer: A NAT gateway placed in a public subnet

    A NAT gateway placed in a public subnet lets instances in the private subnet start outbound connections, such as downloading OS updates, while dropping any connection that starts from the internet. 'An internet gateway attached to a private subnet' would make that subnet public and open the VMs to inbound traffic, the opposite of the goal. 'A transit gateway placed in a public subnet' links VPCs and on-premises networks to each other and provides no internet egress by itself. 'A VPN gateway attached to a private subnet' carries encrypted tunnels to another site, not general internet access.

  37. Several hospitals that must follow the same healthcare regulations agree to jointly build and share a single cloud environment whose costs and governance are split among them, while outside organizations have no access. Which cloud deployment model does this describe?

    • A.Private cloud
    • B.Hybrid cloud
    • C.Virtual private cloud
    • D.Community cloud
    Show answerHide answer

    Correct answer: Community cloud

    Community cloud is correct: several organizations with shared concerns, here the same healthcare regulations, jointly own and govern one environment, split its costs and keep outsiders out. A private cloud serves a single organization, not a group of hospitals. A hybrid cloud links two or more distinct models, such as private and public, rather than describing who shares one environment. A virtual private cloud is an isolated network that one customer carves out of a public provider, so it is not jointly governed by several members.

  38. During migration planning, an organization decides that one application will stay on its current on-premises servers unchanged for now because moving it offers little benefit and carries compliance risk, although it may be revisited later. Which of the 6 R application migration strategies is this?

    • A.Retire
    • B.Repurchase
    • C.Replatform
    • D.Retain
    Show answerHide answer

    Correct answer: Retain

    Retain is the strategy for an application that deliberately stays where it is, because moving it brings little benefit or carries compliance risk, with the option to revisit it later. Retire would shut the application down, but this one keeps running. Repurchase replaces it with a commercial SaaS product, and replatform moves it to the cloud with small optimizations; both act on it now.

  39. A team already defines its servers, networks, and load balancers in version-controlled templates. They now want a separate code-driven approach that manages the operating system settings, installed packages, and application settings inside those provisioned servers. Which practice specifically addresses managing that internal configuration?

    • A.Infrastructure as code
    • B.Configuration as code
    • C.Policy as code
    • D.Pipeline as code
    Show answerHide answer

    Correct answer: Configuration as code

    Configuration as code (CaC) manages what runs inside provisioned servers, such as OS settings, packages and application parameters, through version-controlled definitions applied by tools like Ansible, Puppet or Chef. Infrastructure as code is what the team already uses: it provisions the servers, networks and load balancers themselves. Policy as code writes compliance and access rules as code that a policy engine enforces, and pipeline as code defines the CI/CD build and release stages in a file such as a Jenkinsfile; neither sets up a server's internal configuration.

  40. An engineer runs the same infrastructure-as-code template against an environment that is already fully provisioned, and the tool reports that no changes are needed and creates no duplicate resources. Which property of well-written IaC produces this result?

    • A.Idempotency, so applying the same definition repeatedly converges to a single end state
    • B.Immutability, so servers are replaced with fresh instances instead of changing in place
    • C.State locking, so two engineers cannot apply code changes to a stack at once
    • D.Modularity, so one module of code can be reused across numerous environments
    Show answerHide answer

    Correct answer: Idempotency, so applying the same definition repeatedly converges to a single end state

    Idempotency, so applying the same definition repeatedly converges to a single end state, is the property at work: when the environment already matches the template, applying it again changes nothing and creates no duplicates. Immutability, so servers are replaced with fresh instances instead of changing in place, is a real practice, but it explains how updates are rolled out, not why a re-run makes no changes. State locking, so two engineers cannot apply code changes to a stack at once, prevents conflicting runs. Modularity, so one module of code can be reused across numerous environments, is about reuse, not about repeat runs.

  41. A company is provisioning cloud resources for an application that must store customer records only within the country where the customers live to satisfy a legal requirement. Which provisioning decision most directly satisfies this requirement?

    • A.Encrypting the stored records at rest with customer-managed keys for privacy
    • B.Replicating the database to a distant second site for disaster recovery
    • C.Spreading the instances across availability zones for resilience
    • D.Choosing the cloud region located in that country for data residency
    Show answerHide answer

    Correct answer: Choosing the cloud region located in that country for data residency

    'Choosing the cloud region located in that country for data residency' is the decision that meets the law: a region inside the customers' own country keeps the stored records physically within its borders, which is what data residency requires. Encrypting the stored records at rest with customer-managed keys for privacy protects the data but does not control where it lives. Replicating the database to a distant second site for disaster recovery could copy the records outside the country and break the requirement. Spreading the instances across availability zones for resilience improves availability inside one region but says nothing about the legal location of the data.

Security (37)

  1. Which protocol is primarily used for securely transferring files to and from a cloud storage service?

    • A.SFTP
    • B.FTP
    • C.TFTP
    • D.HTTP
    Show answerHide answer

    Correct answer: SFTP

    SFTP, the SSH File Transfer Protocol, moves files over an encrypted SSH session, so credentials and data are protected in transit to and from cloud storage. Plain FTP sends usernames, passwords and file contents in clear text. TFTP is a minimal protocol with no authentication or encryption, used mainly for network boot and device configuration. HTTP carries web traffic unencrypted; only its TLS-protected form, HTTPS, would secure a transfer.

  2. In a cloud environment, what is the primary function of a cloud access security broker CASB?

    • A.To filter packets between trusted and untrusted networks
    • B.To mediate usage between users and service providers
    • C.To tunnel traffic between remote and corporate networks
    • D.To negotiate pricing between buyers and hosting vendors
    Show answerHide answer

    Correct answer: To mediate usage between users and service providers

    To mediate usage between users and service providers is the function of a cloud access security broker (CASB): it sits between an organization's users and its cloud services and enforces security policy such as access control, data protection and visibility into which services are used. To filter packets between trusted and untrusted networks is a network firewall's job. To tunnel traffic between remote and corporate networks describes a VPN. To negotiate pricing between buyers and hosting vendors is the business role of a cloud service broker, which is not a security control.

  3. What is the primary function of a cloud-based Intrusion Detection System (IDS)?

    • A.To drop malicious network traffic inline on detection
    • B.To flag suspicious network traffic for analyst review
    • C.To filter network traffic by port and address rules
    • D.To encrypt network traffic between remote offices
    Show answerHide answer

    Correct answer: To flag suspicious network traffic for analyst review

    To flag suspicious network traffic for analyst review is the function of an intrusion detection system (IDS): it watches traffic, matches it against signatures or normal baselines and raises alerts, but it does not stop the traffic itself. To drop malicious network traffic inline on detection is what an intrusion prevention system (IPS) adds. To filter network traffic by port and address rules is a firewall's function. To encrypt network traffic between remote offices is the job of a VPN.

  4. What is the primary purpose of a Cloud Access Security Broker CASB in cloud computing?

    • A.To give employees single sign-on to cloud services
    • B.To block injection attacks on internet-facing cloud services
    • C.To enforce security policy on traffic to cloud services
    • D.To track the monthly spending on cloud services
    Show answerHide answer

    Correct answer: To enforce security policy on traffic to cloud services

    To enforce security policy on traffic to cloud services is the purpose of a cloud access security broker (CASB): it sits between users and cloud service providers as a policy enforcement point, applying access, data-protection and compliance rules to cloud use. To give employees single sign-on to cloud services is the job of an identity provider. To block injection attacks on internet-facing cloud services describes a web application firewall. To track the monthly spending on cloud services is cost management, not security brokering.

  5. In cloud computing, what does the term "data sovereignty" refer to?

    • A.Physical location of stored data
    • B.Legal jurisdiction over data
    • C.Custodial control over data
    • D.Sensitivity labeling of stored data
    Show answerHide answer

    Correct answer: Legal jurisdiction over data

    Legal jurisdiction over data is what data sovereignty means: data is subject to the laws and regulations of the country where it is collected or stored, which matters when a cloud provider keeps copies in several countries. Physical location of stored data is data residency, the fact that sovereignty depends on, not sovereignty itself. Custodial control over data is the day-to-day responsibility of a data custodian. Sensitivity labeling of stored data is data classification.

  6. Which cloud security concept involves distributing data across multiple storage devices to mitigate the risk of data loss?

    • A.Data striping
    • B.Data encryption
    • C.Data dispersion
    • D.Data masking
    Show answerHide answer

    Correct answer: Data dispersion

    Data dispersion spreads data, often as encoded fragments, across multiple storage devices and locations so that losing any one device or site does not lose the data. Data striping also spreads blocks across disks, but plain striping as in RAID 0 adds no redundancy, so one failed disk loses everything. Data encryption keeps data unreadable without the key, which protects confidentiality but adds no extra copies. Data masking hides sensitive values from viewers and does nothing to prevent loss.

  7. In the context of cloud security, what is the primary purpose of tokenization?

    • A.To change sensitive data into reversible ciphertext
    • B.To reduce user passwords to a fixed-length digest
    • C.To issue signed session tokens for user logins
    • D.To replace sensitive data with non-sensitive values
    Show answerHide answer

    Correct answer: To replace sensitive data with non-sensitive values

    To replace sensitive data with non-sensitive values is tokenization: a card number or other sensitive element is replaced by a token with no exploitable meaning, and the real value stays in a secured token vault. To change sensitive data into reversible ciphertext is encryption, where anyone able to decrypt it can recover the original value. To reduce user passwords to a fixed-length digest is hashing. To issue signed session tokens for user logins is authentication, which uses the word token in a different sense.

  8. In cloud computing, what is a primary security concern of multi-tenancy?

    • A.Data exposure between tenants
    • B.Resource contention among tenants
    • C.Phishing of tenant administrators
    • D.Weak passwords on tenant accounts
    Show answerHide answer

    Correct answer: Data exposure between tenants

    Data exposure between tenants is the main security concern of multi-tenancy: when customers share hosts, storage and networks, weak isolation can let one tenant read or reach another tenant's data. Resource contention among tenants is the noisy-neighbor problem, which hurts performance rather than security. Phishing of tenant administrators and weak passwords on tenant accounts are security risks in any environment, single-tenant or shared, so they are not concerns created by multi-tenancy.

  9. What is the main function of a Web Application Firewall (WAF) in cloud environments?

    • A.To deny DNS lookups for known phishing domains
    • B.To filter inbound IP packets by port number
    • C.To encrypt IP traffic among branch office sites
    • D.To block malicious HTTP requests to hosted apps
    Show answerHide answer

    Correct answer: To block malicious HTTP requests to hosted apps

    To block malicious HTTP requests to hosted apps is the job of a web application firewall (WAF): it filters and monitors HTTP and HTTPS traffic at the application layer, stopping attacks such as SQL injection and cross-site scripting before they reach the web application. To deny DNS lookups for known phishing domains is DNS filtering. To filter inbound IP packets by port number is a network firewall or security group working at layers 3 and 4. To encrypt IP traffic among branch office sites is an IPsec VPN.

  10. Which of the following best describes the purpose of "Cloud Security Posture Management" (CSPM)?

    • A.Automating the detection and repair of misconfigurations
    • B.Centralizing the collection and analysis of logs
    • C.Protecting running workloads from malware and exploits
    • D.Supplying threat intelligence and indicators from external feeds
    Show answerHide answer

    Correct answer: Automating the detection and repair of misconfigurations

    Automating the detection and repair of misconfigurations is the purpose of Cloud Security Posture Management (CSPM): it continuously checks cloud accounts against security and compliance baselines and flags or fixes risky settings across IaaS, PaaS and SaaS. Centralizing the collection and analysis of logs is the role of a SIEM. Protecting running workloads from malware and exploits is the job of a cloud workload protection platform (CWPP). Supplying threat intelligence and indicators from external feeds describes a threat intelligence service.

  11. In a cloud environment, what is the main purpose of using "Cloud Workload Protection Platforms" (CWPP)?

    • A.To audit accounts and settings against compliance baselines
    • B.To host code and databases for application deployment
    • C.To gather logs and events for central analysis
    • D.To shield hosts and containers against runtime attacks
    Show answerHide answer

    Correct answer: To shield hosts and containers against runtime attacks

    To shield hosts and containers against runtime attacks is the purpose of a cloud workload protection platform (CWPP): it protects the workloads themselves, such as virtual machines, containers and serverless functions, with vulnerability scanning, hardening, application control and runtime threat detection. To audit accounts and settings against compliance baselines is cloud security posture management (CSPM). To host code and databases for application deployment is what a platform as a service offers; the word platform in the name does not make a CWPP one. To gather logs and events for central analysis is the role of a SIEM.

  12. Which encryption method in cloud computing involves encrypting data before it leaves the client's network and enters the cloud provider's environment?

    • A.At-rest encryption
    • B.Server-side encryption
    • C.Client-side encryption
    • D.In-transit encryption
    Show answerHide answer

    Correct answer: Client-side encryption

    Client-side encryption encrypts data on the customer's own systems before it is sent, so it is already ciphertext when it reaches the provider, and the keys can stay under customer control. Server-side encryption happens after the provider receives the data. In-transit encryption, such as TLS, protects data only while it moves across the network and leaves it readable at each end. At-rest encryption protects stored data on the provider's disks, again applied after the data arrives.

  13. In cloud computing, what is the main purpose of Identity and Access Management (IAM)?

    • A.To authenticate users and control their privileges
    • B.To generate encryption keys and rotate them
    • C.To inspect packets and drop unwanted traffic
    • D.To catalog sensitive data and prevent exfiltration
    Show answerHide answer

    Correct answer: To authenticate users and control their privileges

    To authenticate users and control their privileges is the purpose of identity and access management (IAM): it verifies who each user or service is and then controls which resources it may use, through accounts, roles and permission policies. To generate encryption keys and rotate them is the job of a key management service. To inspect packets and drop unwanted traffic is a firewall's function. To catalog sensitive data and prevent exfiltration describes data loss prevention (DLP).

  14. What type of cloud service model is particularly susceptible to hypervisor attacks?

    • A.Platform as a Service (PaaS)
    • B.Business Process as a Service (BPaaS)
    • C.Software as a Service (SaaS)
    • D.Infrastructure as a Service (IaaS)
    Show answerHide answer

    Correct answer: Infrastructure as a Service (IaaS)

    Infrastructure as a Service (IaaS) is the model most exposed to hypervisor attacks because customers run their own virtual machines directly on the provider's shared hypervisors, so a VM-escape flaw could let one tenant's machine reach the host or other tenants' machines. Platform as a Service (PaaS) gives customers no direct access to virtual machines; the hypervisor stays hidden behind the provider's platform. Software as a Service (SaaS) hides it further, behind a finished application. Business Process as a Service (BPaaS) delivers a complete outsourced business process such as payroll, far removed from any hypervisor.

  15. Which cloud security threat involves the unauthorized exploitation of valid cloud service features?

    • A.SQL injection attacks
    • B.DDoS attacks
    • C.API vulnerabilities
    • D.VM escape
    Show answerHide answer

    Correct answer: API vulnerabilities

    API vulnerabilities are the threat described: attackers call the same legitimate APIs that cloud services expose for management and integration, exploiting weak authentication, excessive permissions or poor input checks to perform actions they are not authorized to perform. SQL injection attacks target a database through unsanitized application input rather than through a cloud service's own features. DDoS attacks flood a target with traffic to exhaust its capacity instead of exploiting service functions. VM escape is an attack that breaks out of a virtual machine into the hypervisor.

  16. In cloud security, what does the Shared Responsibility Model imply?

    • A.The provider and customer split security tasks
    • B.The provider handles security for customer data
    • C.The customer handles security for provider hosts
    • D.The provider and customer hand security to a broker
    Show answerHide answer

    Correct answer: The provider and customer split security tasks

    The provider and customer split security tasks under the shared responsibility model: the provider secures the underlying cloud, such as facilities, hardware and the virtualization layer, while the customer secures what it puts in the cloud, such as data, identities and configurations, with the exact split depending on the service model. The provider handles security for customer data is a common misreading; customers remain responsible for their own data and access in every model. The customer handles security for provider hosts is the reverse error, since customers do not secure the provider's data centers or virtualization layer. The provider and customer hand security to a broker is wrong because a broker or other third party can help but does not take over either side's accountability.

  17. What is the primary function of Secure Sockets Layer (SSL) in cloud computing?

    • A.To issue certificates to web servers and domains
    • B.To open shells and run commands on cloud servers
    • C.To sign and publish DNS records on name servers
    • D.To encrypt data sent between clients and servers
    Show answerHide answer

    Correct answer: To encrypt data sent between clients and servers

    To encrypt data sent between clients and servers is the function of Secure Sockets Layer (SSL), now superseded by TLS: it creates an encrypted session so data crossing the Internet cannot be read or altered in transit. To issue certificates to web servers and domains is the job of a certificate authority; SSL uses those certificates but does not issue them. To open shells and run commands on cloud servers is what SSH does. To sign and publish DNS records on name servers is DNSSEC, which proves records are authentic but does not encrypt them.

  18. Which cloud security measure is specifically designed to protect data by converting it into unreadable code that requires a key to decrypt?

    • A.Encryption
    • B.Tokenization
    • C.Hashing
    • D.Data masking
    Show answerHide answer

    Correct answer: Encryption

    Encryption is right: it turns readable plaintext into ciphertext that can be turned back only with the matching key. Tokenization swaps the value for a random token and keeps the original in a vault, so nothing is decrypted; hashing is a one-way function that no key can reverse; data masking permanently hides or scrambles values for testing or display and is not decrypted either.

  19. In cloud computing, what is the main purpose of implementing a Zero Trust security model?

    • A.To trust users once they are inside the network perimeter
    • B.To demand fresh verification for each access request
    • C.To place one strong firewall at the edge of the network
    • D.To let users sign in once and reach their applications
    Show answerHide answer

    Correct answer: To demand fresh verification for each access request

    Zero Trust exists to demand fresh verification for each access request: no user, device or network location is trusted by default (NIST SP 800-207). The option 'to trust users once they are inside the network perimeter' is the older castle-and-moat model that Zero Trust replaces; 'to place one strong firewall at the edge of the network' is the same perimeter thinking; 'to let users sign in once and reach their applications' describes single sign-on, a convenience that Zero Trust still re-checks on each request.

  20. Which of the following best describes the function of Cloud Service Level Agreements (SLAs) in terms of security?

    • A.Scanning hosted systems for security flaws each week
    • B.Proving security controls work via independent audits
    • C.Defining security procedures the provider will follow
    • D.Collecting security event logs for real-time analysis
    Show answerHide answer

    Correct answer: Defining security procedures the provider will follow

    On the security side, an SLA's job is defining security procedures the provider will follow, so the customer has written commitments to hold the provider to. Scanning hosted systems for security flaws each week is vulnerability scanning, an operational task rather than a contract term; proving security controls work via independent audits is what an attestation report such as SOC 2 does; collecting security event logs for real-time analysis is the work of a SIEM tool.

  21. In the context of cloud security, what is the primary role of a Virtual Private Network (VPN)?

    • A.To lease a dedicated private line instead of the Internet
    • B.To carry traffic in an encrypted tunnel over the Internet
    • C.To cache web content at CDN edge locations
    • D.To spread web requests across several HTTP servers
    Show answerHide answer

    Correct answer: To carry traffic in an encrypted tunnel over the Internet

    A VPN's role is to carry traffic in an encrypted tunnel over the Internet, so data crossing the public network cannot be read or altered in transit. 'To lease a dedicated private line instead of the Internet' describes a direct connection service such as AWS Direct Connect or Azure ExpressRoute, not a VPN; 'to cache web content at CDN edge locations' is a content delivery network; 'to spread web requests across several HTTP servers' is load balancing.

  22. Which cloud maintenance activity is specifically focused on identifying and addressing vulnerabilities within the cloud infrastructure?

    • A.Load balancing
    • B.Load testing
    • C.Vulnerability disclosure
    • D.Vulnerability scanning
    Show answerHide answer

    Correct answer: Vulnerability scanning

    Vulnerability scanning is the routine maintenance task that probes systems for known weaknesses, such as missing patches and unsafe settings, so they can be fixed. Load balancing spreads traffic across servers; load testing measures how a system performs under heavy demand; vulnerability disclosure is the process of reporting a flaw someone has already found to the vendor or the public, not a task that searches your own infrastructure.

  23. What tool or technique in cloud management is most effective for detecting unauthorized access to cloud resources?

    • A.Intrusion detection systems (IDS)
    • B.Network address translation (NAT)
    • C.Hardware security modules (HSM)
    • D.Multifactor authentication (MFA)
    Show answerHide answer

    Correct answer: Intrusion detection systems (IDS)

    Intrusion detection systems (IDS) watch network and host activity for signs of unauthorized access and raise alerts when they find them. Network address translation (NAT) rewrites addresses between private and public networks and detects nothing; hardware security modules (HSM) store and use encryption keys; multifactor authentication (MFA) makes unauthorized sign-ins harder to achieve but does not detect access that is already happening.

  24. A cloud engineer is hardening logins for an administrative console. Beyond a username and password, the policy now requires a one-time code from an authenticator app. Which security control is being implemented?

    • A.Privileged access management
    • B.Role-based access control
    • C.Certificate-based authentication
    • D.Multi-factor authentication
    Show answerHide answer

    Correct answer: Multi-factor authentication

    Multi-factor authentication is being implemented: the password is something the administrator knows and the authenticator app code is something they have, so a stolen password alone no longer opens the console. Privileged access management governs and records the use of powerful admin accounts but does not by itself add a second factor at login. Role-based access control decides what a signed-in user may do, not how they prove who they are. Certificate-based authentication replaces the login with a client certificate rather than adding a one-time code to the password.

  25. A public-facing web application in the cloud is being targeted with SQL injection and cross-site scripting attempts embedded in HTTP requests. Which control is purpose-built to inspect and filter this Layer 7 web traffic?

    • A.A network intrusion detection system
    • B.A secure web gateway
    • C.A web application firewall
    • D.A network access control list
    Show answerHide answer

    Correct answer: A web application firewall

    A web application firewall is built for this: it works at Layer 7, reads each HTTP request and blocks patterns such as SQL injection and cross-site scripting before they reach the application. A network intrusion detection system can spot the attack, but it only raises alerts and does not filter the requests. A secure web gateway filters the outbound browsing of an organization's own users rather than protecting a public application from inbound requests. A network access control list allows or denies traffic by IP address, port and protocol and cannot read what is inside a web request.

  26. In a cloud environment, which term describes the framework of policies and technologies that ensures the right individuals and services have the appropriate access to resources?

    • A.Security information and event management
    • B.Audit and accountability
    • C.Identity and access management
    • D.Security orchestration, automation and response
    Show answerHide answer

    Correct answer: Identity and access management

    Identity and access management (IAM) is the framework that defines the identities of users and services and controls which resources each may reach, through authentication and authorization. Security information and event management collects and correlates logs to detect threats, but it does not grant or limit access. Audit and accountability records and reviews what identities did after the fact rather than deciding who may use which resource. Security orchestration, automation and response runs playbooks that react to incidents once they are detected, and it grants no access itself.

  27. A cloud team grants permissions by assigning users to predefined job functions such as DatabaseAdmin or Auditor, and each function carries a fixed set of permissions. Which access control model is in use?

    • A.Mandatory access control
    • B.Discretionary access control
    • C.Role-based access control
    • D.Attribute-based access control
    Show answerHide answer

    Correct answer: Role-based access control

    Role-based access control (RBAC) is in use: permissions are attached to roles that match job functions, such as DatabaseAdmin or Auditor, and users gain them by being placed in a role. Mandatory access control grants access by comparing classification labels set by a central authority. Discretionary access control lets each resource owner decide who may use it. Attribute-based access control evaluates attributes of the user, resource and context on each request rather than a fixed set tied to a job role.

  28. A security architecture removes the assumption that traffic inside the corporate network perimeter is trustworthy and instead verifies every request based on identity and device posture, regardless of network location. Which model does this describe?

    • A.Defense in depth
    • B.Perimeter-based security
    • C.Microsegmentation
    • D.Zero trust
    Show answerHide answer

    Correct answer: Zero trust

    Zero trust is described: it follows 'never trust, always verify', treats no user, device or network location as trusted by default, and checks identity and device posture on every request. Defense in depth stacks several layers of controls but can still treat the internal network as trusted. Perimeter-based security is the castle-and-moat model zero trust replaces, since it trusts anything inside the firewall. Microsegmentation is one technique a zero trust design may use to limit lateral movement, not the overall model of verifying every request.

  29. A cloud engineer must choose between two authorization models. One grants access strictly by assigned roles, while the other evaluates dynamic attributes such as department, time of day, and resource sensitivity at the moment of the request. Which statement correctly distinguishes them?

    • A.ABAC evaluates live attributes on each request, while RBAC grants access through fixed job roles
    • B.ABAC lets each resource owner share access at will, while RBAC enforces central clearance labels
    • C.RBAC evaluates live attributes on each request, while ABAC grants access through fixed job roles
    • D.RBAC lets each resource owner share access at will, while ABAC enforces central clearance labels
    Show answerHide answer

    Correct answer: ABAC evaluates live attributes on each request, while RBAC grants access through fixed job roles

    'ABAC evaluates live attributes on each request, while RBAC grants access through fixed job roles' is the correct distinction: attribute-based access control weighs attributes such as department, time of day and resource sensitivity when the request arrives, while role-based access control ties permissions to static roles. 'RBAC evaluates live attributes on each request, while ABAC grants access through fixed job roles' swaps the two models. 'ABAC lets each resource owner share access at will, while RBAC enforces central clearance labels' and its mirror 'RBAC lets each resource owner share access at will, while ABAC enforces central clearance labels' describe discretionary and mandatory access control, which are neither model.

  30. A cloud network team wants to apply granular firewall policies between individual workloads inside the same subnet so that a compromised server cannot freely reach its neighbors. Which technique achieves this east-west isolation?

    • A.Microsegmentation
    • B.VLAN segmentation
    • C.Network address translation
    • D.Port mirroring
    Show answerHide answer

    Correct answer: Microsegmentation

    Microsegmentation achieves this: it applies fine-grained policies down to the individual workload, so even servers in the same subnet can reach each other only where a rule allows, which limits lateral, east-west movement after a breach. VLAN segmentation separates whole network segments, but workloads that share a segment can still talk freely. Network address translation rewrites addresses at a boundary and does not filter traffic between neighbors. Port mirroring copies traffic to a monitoring tool and blocks nothing.

  31. A compliance requirement states that customer records stored on cloud block volumes and object storage must be unreadable if the underlying disks are physically stolen. Which control directly satisfies this requirement?

    • A.Encryption in transit
    • B.Data loss prevention
    • C.Encryption at rest
    • D.Badge readers at the data center
    Show answerHide answer

    Correct answer: Encryption at rest

    Encryption at rest satisfies the requirement: data on block volumes and object storage is kept as ciphertext, so a stolen disk is unreadable without the decryption key. Encryption in transit protects data moving across networks, not data sitting on a disk. Data loss prevention watches for sensitive data leaving through email, uploads or other channels and does not protect a disk that has been carried away. Badge readers at the data center make theft harder, but once a disk is taken its contents are still readable.

  32. In a cloud virtual network, which construct acts as a virtual, stateful firewall attached to an instance's network interface to control its inbound and outbound traffic?

    • A.A network access control list
    • B.A security group
    • C.A route table
    • D.A virtual private gateway
    Show answerHide answer

    Correct answer: A security group

    A security group is the virtual, stateful firewall attached to an instance's network interface; because it tracks connections, return traffic for an allowed request is let through automatically. A network access control list also filters traffic, but it is stateless and applies to a whole subnet. A route table decides where traffic is sent and filters nothing. A virtual private gateway terminates VPN connections into the virtual network rather than controlling one instance's traffic.

  33. A cloud architect is documenting data protection. One mechanism protects information as it moves between a client and a server over the internet using TLS, and another protects information sitting on storage volumes. Which pairing correctly maps these two states?

    • A.Encryption in transit protects stored data, and encryption at rest secures network traffic
    • B.Both encryption in transit and encryption at rest cover packets as they cross the internet
    • C.Both encryption in transit and encryption at rest cover files as they sit on a server disk
    • D.Encryption in transit protects network traffic, and encryption at rest secures stored data
    Show answerHide answer

    Correct answer: Encryption in transit protects network traffic, and encryption at rest secures stored data

    'Encryption in transit protects network traffic, and encryption at rest secures stored data' is the correct mapping: TLS between client and server is encryption in transit, while ciphertext on volumes, databases and object storage is encryption at rest. 'Encryption in transit protects stored data, and encryption at rest secures network traffic' swaps the two. 'Both encryption in transit and encryption at rest cover packets as they cross the internet' and 'both encryption in transit and encryption at rest cover files as they sit on a server disk' each collapse two different states into one, when a complete design needs both.

  34. A cloud IAM administrator assigns each service account only the specific permissions it needs to perform its function and nothing more. Which security principle is being applied?

    • A.Least privilege
    • B.Separation of duties
    • C.Least functionality
    • D.Rotation of duties
    Show answerHide answer

    Correct answer: Least privilege

    Least privilege is being applied: each identity, including a service account, gets only the permissions its task requires, which shrinks the attack surface if it is compromised. Least functionality is a different principle: it configures a system to offer only essential services, ports and features, rather than limiting what an account may do. Separation of duties splits a sensitive task across several people to prevent fraud or error, and rotation of duties moves staff between roles so misconduct surfaces; neither trims one account's permissions.

  35. Under the cloud shared responsibility model, which party is responsible for the security of the physical data center hardware, including servers, storage devices, and facility access controls?

    • A.The cloud consumer
    • B.The cloud auditor
    • C.The cloud provider
    • D.The cloud carrier
    Show answerHide answer

    Correct answer: The cloud provider

    The cloud provider secures the physical data center, its hardware and facility access: the 'security of the cloud' half of the shared responsibility model. The cloud consumer, the customer, is responsible for 'security in the cloud', such as its data, identities, access settings and, in IaaS, guest OS patching, but never the physical facility. The cloud auditor independently assesses controls and owns none of them. The cloud carrier provides network connectivity and transport between the consumer and the provider.

  36. A cloud engineer must explain why a security group and a network access control list (NACL) behave differently when filtering traffic. Which statement is accurate?

    • A.A security group is stateless and operates at the instance level, while a NACL is stateful and operates at the subnet level
    • B.A security group is stateful and operates at the instance level, while a NACL is stateless and operates at the subnet level
    • C.A security group is stateless and operates at the subnet level, while a NACL is stateful and operates at the instance level
    • D.A security group is stateful and operates at the subnet level, while a NACL is stateless and operates at the instance level
    Show answerHide answer

    Correct answer: A security group is stateful and operates at the instance level, while a NACL is stateless and operates at the subnet level

    'A security group is stateful and operates at the instance level, while a NACL is stateless and operates at the subnet level' is accurate: a security group remembers allowed connections and lets return traffic back automatically, while a network ACL judges every packet on its own and guards a whole subnet. 'A security group is stateless and operates at the instance level, while a NACL is stateful and operates at the subnet level' swaps the statefulness. 'A security group is stateful and operates at the subnet level, while a NACL is stateless and operates at the instance level' swaps where each applies. 'A security group is stateless and operates at the subnet level, while a NACL is stateful and operates at the instance level' swaps both.

  37. A regulatory standard requires that login credentials and form submissions cannot be read if intercepted while traveling between a user's browser and a cloud-hosted web application. Which protection addresses this specific scenario?

    • A.Encryption of data in transit
    • B.Encryption of data at rest
    • C.Hashing of stored credentials
    • D.Masking of stored form fields
    Show answerHide answer

    Correct answer: Encryption of data in transit

    Encryption of data in transit addresses this: TLS wraps the session between the browser and the application, so anyone who intercepts the packets sees only ciphertext. Encryption of data at rest protects information stored on disks and databases, not credentials moving across the network. Hashing of stored credentials protects saved passwords if the database is stolen, but the login still travels in the clear without TLS. Masking of stored form fields hides values from people viewing records and does nothing for the submission in flight.

Operations (33)

  1. What is the purpose of using a cloud service catalog?

    • A.To track deployed cloud services for availability reporting
    • B.To list approved cloud services for user self-provisioning
    • C.To itemize consumed cloud services for departmental invoicing
    • D.To scan active cloud services for compliance auditing
    Show answerHide answer

    Correct answer: To list approved cloud services for user self-provisioning

    To list approved cloud services for user self-provisioning is the purpose of a service catalog: it publishes the standard offerings users may order, so they can provision them on demand within agreed policies. To track deployed cloud services for availability reporting is monitoring. To itemize consumed cloud services for departmental invoicing is chargeback or showback billing. To scan active cloud services for compliance auditing describes posture-management and audit tools rather than a catalog.

  2. In a cloud environment, what is the main purpose of implementing disaster recovery as a service (DRaaS)?

    • A.To resume business operations quickly after a major outage
    • B.To add compute capacity quickly after a traffic surge
    • C.To patch exposed servers quickly after a vendor advisory
    • D.To stop hostile traffic quickly after a detected intrusion
    Show answerHide answer

    Correct answer: To resume business operations quickly after a major outage

    To resume business operations quickly after a major outage is the purpose of disaster recovery as a service (DRaaS): the provider replicates systems and data to its cloud so they can be failed over and restored with minimal downtime, keeping the business running. To add compute capacity quickly after a traffic surge is auto-scaling, a response to load rather than to a disaster. To patch exposed servers quickly after a vendor advisory is patch management. To stop hostile traffic quickly after a detected intrusion is the job of an intrusion prevention system or firewall.

  3. What is the primary function of a cloud management platform (CMP)?

    • A.To correlate logs across providers into alerts
    • B.To encrypt stored data across providers with customer keys
    • C.To replicate virtual machines across providers for failover
    • D.To oversee resources across providers from one console
    Show answerHide answer

    Correct answer: To oversee resources across providers from one console

    To oversee resources across providers from one console is the function of a cloud management platform (CMP): it gives a single, unified interface for provisioning, monitoring, governing and costing services across public and private clouds. To correlate logs across providers into alerts is the job of a SIEM. To encrypt stored data across providers with customer keys describes key management with customer-managed keys. To replicate virtual machines across providers for failover is disaster recovery replication rather than unified management.

  4. Which process is essential for ensuring data integrity during cloud-based backups?

    • A.Data deduplication
    • B.Hash verification
    • C.Incremental backup
    • D.Data compression
    Show answerHide answer

    Correct answer: Hash verification

    Hash verification protects integrity: a hash of the data is taken before the backup and compared with a hash of the stored copy, and any mismatch shows the data changed or was corrupted. Data deduplication removes repeated blocks to save space; an incremental backup copies only what changed since the last backup; data compression shrinks the backup. None of the three proves the copy matches the original.

  5. Which cloud maintenance activity involves updating the underlying physical hardware and software infrastructure components?

    • A.Patch management
    • B.Infrastructure monitoring
    • C.Capacity management
    • D.Infrastructure refresh
    Show answerHide answer

    Correct answer: Infrastructure refresh

    Infrastructure refresh is the planned updating or replacement of aging physical hardware and the software that runs on it. Patch management applies software fixes but does not replace hardware; infrastructure monitoring watches components for faults and performance without changing them; capacity management sizes resources to demand rather than renewing the equipment itself.

  6. In cloud environments, what is the primary purpose of implementing a disaster recovery plan?

    • A.To satisfy data retention rules set by regulators
    • B.To schedule patches inside approved maintenance windows
    • C.To lower monthly spending on compute workloads
    • D.To ensure business continuity during major outages
    Show answerHide answer

    Correct answer: To ensure business continuity during major outages

    A disaster recovery plan exists to ensure business continuity during major outages, setting out how systems and data are restored and who does what. 'To satisfy data retention rules set by regulators' is a compliance goal; 'to schedule patches inside approved maintenance windows' belongs to patch and change management; 'to lower monthly spending on compute workloads' is cost optimization. A recovery plan may touch these, but none is its purpose.

  7. What is the main goal of capacity planning in a cloud environment?

    • A.To have enough resources ready for future demand
    • B.To keep cloud spending under the approved budget
    • C.To restore the service quickly after an outage
    • D.To tune live workloads for faster response times
    Show answerHide answer

    Correct answer: To have enough resources ready for future demand

    Capacity planning aims to have enough resources ready for future demand, using usage trends and forecasts so growth does not outrun the environment. 'To keep cloud spending under the approved budget' is cost management; 'to restore the service quickly after an outage' is disaster recovery; 'to tune live workloads for faster response times' is performance tuning, which works on today's load rather than forecasting tomorrow's.

  8. Which of the following is a key consideration when implementing cloud automation?

    • A.Compatibility of tools with current systems
    • B.Maintenance of the physical server hardware
    • C.Storage of credentials inside each automation script
    • D.Selection of products by license cost alone
    Show answerHide answer

    Correct answer: Compatibility of tools with current systems

    Compatibility of tools with current systems is a key consideration: automation has to work with the platforms, APIs and applications already in place, or it breaks more than it saves. Maintenance of the physical server hardware is the provider's job in the cloud and has nothing to do with automating tasks; storage of credentials inside each automation script is a known bad practice, since secrets belong in a vault; selection of products by license cost alone ignores fit, which is exactly what makes automation fail.

  9. In cloud computing, what is the primary purpose of implementing change management?

    • A.To track software licenses and their renewals
    • B.To guard data center entrances and server racks
    • C.To control how new services and updates go live
    • D.To log and resolve issues sent to the help desk
    Show answerHide answer

    Correct answer: To control how new services and updates go live

    Change management exists to control how new services and updates go live, so each change is requested, reviewed, approved, scheduled and rolled back if needed. 'To track software licenses and their renewals' is license management; 'to guard data center entrances and server racks' is physical security; 'to log and resolve issues sent to the help desk' is the service desk's incident and request handling.

  10. In cloud computing, what is the main purpose of implementing automated scaling?

    • A.To spread incoming requests evenly across servers
    • B.To cut manual work in allocating resources
    • C.To remove duplicate data from storage
    • D.To fail over to another region after an outage
    Show answerHide answer

    Correct answer: To cut manual work in allocating resources

    Automated scaling is there to cut manual work in allocating resources: rules or metrics add and remove capacity on their own instead of an administrator doing it. 'To spread incoming requests evenly across servers' is load balancing, which shares traffic but does not change the number of resources; 'to remove duplicate data from storage' is deduplication; 'to fail over to another region after an outage' is disaster recovery.

  11. Which activity is crucial for maintaining data accuracy and consistency across multiple cloud storage locations?

    • A.Data compression
    • B.Data archiving
    • C.Data replication
    • D.Data tiering
    Show answerHide answer

    Correct answer: Data replication

    Data replication copies data to each storage location and keeps the copies synchronized, which is what keeps them accurate and consistent. Data compression encodes data in fewer bits to save space; data archiving moves old data to low-cost long-term storage; data tiering shifts data between faster and cheaper tiers by how often it is used. None of these keeps several live copies in step.

  12. In the context of cloud maintenance, what is the primary function of configuration management tools?

    • A.To inspect network traffic and block suspicious packets
    • B.To store user identities and issue access tokens
    • C.To copy data offsite and restore it after failures
    • D.To track system settings and automate changes
    Show answerHide answer

    Correct answer: To track system settings and automate changes

    Configuration management tools exist to track system settings and automate changes, keeping servers in a known, consistent state and applying updates the same way everywhere. 'To inspect network traffic and block suspicious packets' describes an intrusion prevention system; 'to store user identities and issue access tokens' is an identity provider; 'to copy data offsite and restore it after failures' is backup and recovery.

  13. What is the main advantage of implementing predictive maintenance in a cloud environment?

    • A.Restoring service quickly after a failure occurs
    • B.Fixing likely faults before systems fail
    • C.Lowering the cost of each stored gigabyte
    • D.Freeing floor space in the data center
    Show answerHide answer

    Correct answer: Fixing likely faults before systems fail

    Predictive maintenance uses telemetry and trend analysis for fixing likely faults before systems fail, so problems are handled on a schedule instead of as outages. Restoring service quickly after a failure occurs is reactive recovery, the opposite timing; lowering the cost of each stored gigabyte comes from storage tiering or lifecycle rules; freeing floor space in the data center comes from consolidation, and in the cloud the provider owns the floor anyway.

  14. In cloud environments, what is the purpose of log file analysis?

    • A.To install and test newer software versions
    • B.To surface security and performance issues
    • C.To verify passwords and grant user sign-ins
    • D.To size and assign compute resources to workloads
    Show answerHide answer

    Correct answer: To surface security and performance issues

    Log file analysis is used to surface security and performance issues by reviewing what systems recorded: failed logins, errors, slow requests and unusual patterns. 'To install and test newer software versions' is patching and release work; 'to verify passwords and grant user sign-ins' is the job of an authentication service; 'to size and assign compute resources to workloads' is resource allocation. Logs may later inform those tasks, but analyzing them does not perform them.

  15. Which maintenance process is involved in updating cloud applications to newer versions?

    • A.Patch management
    • B.Capacity management
    • C.Cost management
    • D.Incident management
    Show answerHide answer

    Correct answer: Patch management

    Patch management is the process of finding, testing and applying updates that bring cloud applications to newer versions. Capacity management sizes resources against demand; cost management tracks and controls spending; incident management restores service after an unplanned disruption. None of the three installs new versions.

  16. What is the primary goal of disaster recovery testing in a cloud environment?

    • A.To confirm the plan restores service as intended
    • B.To find exploitable flaws in the public website
    • C.To measure the peak load the application handles
    • D.To check the nightly backups finish on schedule
    Show answerHide answer

    Correct answer: To confirm the plan restores service as intended

    Disaster recovery testing is done to confirm the plan restores service as intended, for example through tabletop exercises or failover drills checked against the RTO and RPO. 'To find exploitable flaws in the public website' is penetration testing; 'to measure the peak load the application handles' is load testing; 'to check the nightly backups finish on schedule' is routine backup monitoring, which shows a job ran but not that the business can actually recover.

  17. Which process in cloud maintenance involves checking and managing the licenses of software used in the cloud environment?

    • A.Resource isolation
    • B.License mobility
    • C.Resource tagging
    • D.License management
    Show answerHide answer

    Correct answer: License management

    License management is the process of tracking which software licenses are in use, checking them against what was bought and keeping the environment compliant. Resource isolation keeps each tenant's workloads and data separate from other tenants; license mobility is a vendor benefit that lets existing licenses move to cloud servers, not a checking process; resource tagging attaches labels to resources for organization and cost tracking.

  18. Which cloud management task is primarily concerned with ensuring compliance with legal and regulatory standards?

    • A.Capacity management
    • B.Budgeting
    • C.Governance
    • D.Security management
    Show answerHide answer

    Correct answer: Governance

    Governance is the management task that sets policies and controls so cloud use stays within legal, regulatory and internal rules. Capacity management sizes resources to expected demand; budgeting sets and tracks spending limits; security management protects systems and data from threats. Security supports compliance, but setting and enforcing those rules is governance.

  19. Which of the following best describes 'chargeback' in cloud computing?

    • A.Reporting usage costs to teams without charging them
    • B.Billing consumers for their actual usage
    • C.Refunding customers for capacity left unused
    • D.Charging a flat monthly subscription fee
    Show answerHide answer

    Correct answer: Billing consumers for their actual usage

    Chargeback means billing consumers for their actual usage: each department, project or client pays for its own measured consumption. Reporting usage costs to teams without charging them is showback, which reports costs but sends no bill; refunding customers for capacity left unused is a billing credit, not chargeback; charging a flat monthly subscription fee ignores consumption altogether.

  20. In cloud management, what is the primary function of Configuration Management Databases CMDB?

    • A.Assigning roles and privileges to users
    • B.Cataloging assets and their current settings
    • C.Tracking cloud spending and trimming waste
    • D.Capturing network packets and flagging anomalies
    Show answerHide answer

    Correct answer: Cataloging assets and their current settings

    A CMDB's main job is cataloging assets and their current settings: each hardware and software configuration item, its configuration and how it relates to other items. Assigning roles and privileges to users is identity and access management; tracking cloud spending and trimming waste is cost optimization; capturing network packets and flagging anomalies is network monitoring or intrusion detection.

  21. In cloud management, which process involves identifying and correcting cloud resource issues before they affect service?

    • A.Corrective maintenance
    • B.Incident response planning
    • C.Proactive maintenance
    • D.Disaster recovery planning
    Show answerHide answer

    Correct answer: Proactive maintenance

    Proactive maintenance finds and fixes resource problems before they reach users, through regular checks, trend monitoring and early fixes. Corrective maintenance repairs a fault after it has already appeared; incident response planning prepares the team to contain and recover from an incident once it has occurred; disaster recovery planning sets out how to restore service after a major event rather than heading off issues.

  22. In cloud computing, what is the primary purpose of a 'service level agreement' SLA?

    • A.To list the prices and billing models of each service
    • B.To set the legal terms and conditions of service contracts
    • C.To explain the hardware specs and features of each service
    • D.To record the target uptime and performance of the service
    Show answerHide answer

    Correct answer: To record the target uptime and performance of the service

    An SLA exists to record the target uptime and performance of the service, such as availability percentages and response times, along with remedies when they are missed. 'To list the prices and billing models of each service' is the pricing schedule; 'to set the legal terms and conditions of service contracts' is the master service agreement that the SLA sits under; 'to explain the hardware specs and features of each service' is product documentation.

  23. Which cloud management function involves monitoring resource usage to control costs and optimize efficiency?

    • A.Load balancing
    • B.Resource tagging
    • C.Capacity planning
    • D.Backup scheduling
    Show answerHide answer

    Correct answer: Capacity planning

    Capacity planning is the answer: it tracks how resources are being used and forecasts what will be needed, so the environment is neither starved nor paying for idle capacity. Load balancing spreads traffic across servers but does not plan resource levels; resource tagging labels resources so costs can be grouped, without monitoring usage itself; backup scheduling sets when data is copied for recovery and has nothing to do with sizing.

  24. In cloud management, what is 'resource tagging' primarily used for?

    • A.Attaching metadata labels to organize resources
    • B.Finding idle resources to switch off
    • C.Encrypting the data held in storage resources
    • D.Monitoring the traffic across network resources
    Show answerHide answer

    Correct answer: Attaching metadata labels to organize resources

    Resource tagging means attaching metadata labels to organize resources, for example by owner, project or environment, which then supports searching, automation and cost reports. Finding idle resources to switch off is rightsizing, done from usage metrics; encrypting the data held in storage resources is data protection; monitoring the traffic across network resources is network monitoring. Tags may feed those tasks but do not perform them.

  25. Which aspect of cloud management involves ensuring that cloud deployments adhere to relevant industry standards and legislation?

    • A.Compliance management
    • B.Cost management
    • C.Continuity management
    • D.Change management
    Show answerHide answer

    Correct answer: Compliance management

    Compliance management checks that cloud deployments meet the industry standards and laws that apply to them, such as PCI DSS, HIPAA or GDPR, and keeps the evidence. Cost management controls spending; continuity management plans how services keep running through a major disruption; change management controls how changes are approved and released. None of these is aimed at meeting standards and legislation.

  26. In cloud management, what is the primary function of 'cost allocation tags'?

    • A.To keep sensitive data encrypted and backed up
    • B.To set which subnets allow and deny traffic
    • C.To assign charges to teams and projects
    • D.To measure service response times and uptime
    Show answerHide answer

    Correct answer: To assign charges to teams and projects

    Cost allocation tags exist to assign charges to teams and projects, so the bill can be broken down by business unit, application or environment. 'To keep sensitive data encrypted and backed up' is data protection; 'to set which subnets allow and deny traffic' is the job of network access control lists; 'to measure service response times and uptime' is performance monitoring.

  27. Which cloud management activity involves regular testing and updating of disaster recovery plans?

    • A.Business impact analysis
    • B.License compliance auditing
    • C.Service continuity planning
    • D.Performance benchmarking
    Show answerHide answer

    Correct answer: Service continuity planning

    Service continuity planning develops the disaster recovery and continuity plans and keeps them current through regular tests and updates. Business impact analysis identifies which functions matter most and how long they can be down, but it does not test or update the plans; license compliance auditing checks software use against the license terms; performance benchmarking compares speed and throughput against a baseline.

  28. A cloud engineer is documenting recovery targets for a production database. The business states it can tolerate losing at most 15 minutes of transactions, but it can accept up to 2 hours of downtime while the system is rebuilt. Which pair of values correctly maps these requirements to the recovery objectives?

    • A.RPO = 2 hours, RTO = 15 minutes
    • B.RPO = 15 minutes, RTO = 15 minutes
    • C.RPO = 2 hours, RTO = 2 hours
    • D.RPO = 15 minutes, RTO = 2 hours
    Show answerHide answer

    Correct answer: RPO = 15 minutes, RTO = 2 hours

    RPO = 15 minutes, RTO = 2 hours is the correct mapping. The recovery point objective is the most data the business can lose, measured back in time, so the 15-minute loss tolerance is the RPO; the recovery time objective is how long service may be down, so the 2-hour rebuild window is the RTO. 'RPO = 2 hours, RTO = 15 minutes' swaps the two objectives, the classic trap. 'RPO = 15 minutes, RTO = 15 minutes' applies the data-loss figure to downtime as well, and 'RPO = 2 hours, RTO = 2 hours' applies the downtime figure to data loss as well.

  29. An organization is choosing a disaster recovery facility. It wants the lowest ongoing cost and is willing to accept the longest recovery time, since it can rebuild and reload data after a regional outage. Which recovery site type best fits this requirement?

    • A.Warm site
    • B.Hot site
    • C.Mirrored site
    • D.Cold site
    Show answerHide answer

    Correct answer: Cold site

    A cold site fits: it provides space, power and connectivity with little or no installed equipment or current data, so it is the cheapest to keep but the slowest to bring online. A warm site keeps some hardware and periodically updated data, recovering faster at a higher running cost. A hot site is fully equipped and nearly current, and a mirrored site runs a complete live duplicate; both recover quickly but cost the most, which contradicts the goal.

  30. A cloud engineer must protect a transactional financial database where any committed write must survive a site failure with zero data loss, even though the two sites are within the same metro region. Which replication approach should be configured?

    • A.Asynchronous replication
    • B.Snapshot replication
    • C.Cross-region replication
    • D.Synchronous replication
    Show answerHide answer

    Correct answer: Synchronous replication

    Synchronous replication is required: a write is acknowledged only after both the primary and the replica have committed it, so a site failure loses no committed data (an RPO of zero). The added latency is acceptable because both sites sit in the same metro region. Asynchronous replication acknowledges the write first and copies it afterward, so the newest transactions can be lost. Snapshot replication copies point-in-time images on a schedule, leaving a gap between snapshots, and cross-region replication is normally asynchronous and spans distant regions, so it also allows some loss.

  31. A cloud team runs a full backup every Sunday. On Monday through Saturday they want each nightly job to copy only the data that changed since the most recent backup of any kind, so each nightly backup runs as fast as possible and uses the least storage. To restore Thursday's data they would then need Sunday's full backup plus every nightly job from Monday through Thursday. Which backup type is being scheduled on the weeknights?

    • A.Incremental backup, because it copies only the data changed since the last backup of any kind
    • B.Differential backup, because it copies the cumulative changes made since the most recent full backup
    • C.Synthetic full backup, because it merges the earlier backups into one new full image
    • D.Mirror backup, because it keeps only an exact replica of the source and drops deleted files
    Show answerHide answer

    Correct answer: Incremental backup, because it copies only the data changed since the last backup of any kind

    Incremental backup, because it copies only the data changed since the last backup of any kind, is what runs on the weeknights: each job is the smallest and fastest, which is why a Thursday restore needs Sunday's full plus every incremental from Monday to Thursday. Differential backup, because it copies the cumulative changes made since the most recent full backup, is true of differentials, but they grow each night and a restore needs only the full plus the latest one. Synthetic full backup, because it merges the earlier backups into one new full image, describes building a full copy without reading the source, not a nightly change job. Mirror backup, because it keeps only an exact replica of the source and drops deleted files, describes a live one-to-one copy with no restore chain.

  32. After a ransomware incident, a cloud operations lead discovers that nightly backups had been running successfully for months but the most recent attempt to recover a database from them failed because the backup files were silently corrupted. Which operational practice would have caught this problem before it was needed in an emergency?

    • A.Periodic restore testing, where a backup is restored on schedule for verification
    • B.Backup job alerting, where a failed nightly backup job pages the engineer on call
    • C.Immutable backup storage, where copies cannot be changed or deleted during a lock
    • D.Longer backup retention, where older copies stay available for a full extra year
    Show answerHide answer

    Correct answer: Periodic restore testing, where a backup is restored on schedule for verification

    Periodic restore testing, where a backup is restored on schedule for verification, would have caught it: a backup is only proven good once it has been restored and checked, and a job reporting success says nothing about whether its files are readable. Backup job alerting, where a failed nightly backup job pages the engineer on call, only fires when a job fails, and these jobs succeeded. Immutable backup storage, where copies cannot be changed or deleted during a lock, stops ransomware from altering backups but preserves files that were already corrupt. Longer backup retention, where older copies stay available for a full extra year, only keeps more copies that may be just as damaged.

  33. A cloud engineer is configuring an autoscaling group for a web tier. The requirement is that the group should add instances when sustained CPU utilization rises above 70 percent and remove instances when it falls below 30 percent, with the actions driven automatically by the monitoring system. What is the engineer defining when they set these 70 percent and 30 percent values?

    • A.Cooldown periods, the wait times that block more scale events after one finishes
    • B.Scaling thresholds, the metric levels that trigger scale-in and scale-out events
    • C.Capacity limits, the minimum and maximum sizes that the group must stay between
    • D.Warm-up periods, the boot times that pass before new nodes count toward averages
    Show answerHide answer

    Correct answer: Scaling thresholds, the metric levels that trigger scale-in and scale-out events

    Scaling thresholds, the metric levels that trigger scale-in and scale-out events, are what the 70 percent and 30 percent values define: when CPU stays above or below them, the monitoring system tells the group to add or remove instances. Cooldown periods, the wait times that block more scale events after one finishes, stop the group from overreacting but do not set the trigger points. Capacity limits, the minimum and maximum sizes that the group must stay between, cap how far it scales rather than when. Warm-up periods, the boot times that pass before new nodes count toward averages, keep a booting instance from skewing the average.

Troubleshooting (23)

  1. In cloud computing, what is the most likely cause of increased latency in data transfer between a user and an application hosted in the cloud?

    • A.Disk fragmentation
    • B.Memory exhaustion
    • C.CPU throttling
    • D.Network congestion
    Show answerHide answer

    Correct answer: Network congestion

    Network congestion is the most likely cause: when links between the user and the application are overloaded, packets queue or are dropped, and transfer latency climbs. Disk fragmentation slows storage reads on the server rather than data in transit; memory exhaustion causes paging and slow processing inside the host; CPU throttling slows computation. Each can lengthen processing time, but none delays the data while it crosses the network.

  2. Which tool or technique is most effective for diagnosing packet loss in a cloud network environment?

    • A.Nslookup
    • B.Traceroute
    • C.Port scanning
    • D.Ipconfig
    Show answerHide answer

    Correct answer: Traceroute

    Traceroute is the right tool: it lists every hop between the source and the destination, so the hop where replies start to drop shows where packets are being lost. Nslookup only queries DNS records and says nothing about the network path. Port scanning shows which ports on a host are open or filtered, not whether packets are being dropped in transit. Ipconfig displays the local machine's IP settings and cannot see beyond the host.

  3. What is the first step in troubleshooting a cloud service that is not performing as expected?

    • A.Consulting the cloud service provider's documentation
    • B.Rebooting the server that hosts the affected application
    • C.Checking the physical network connections to the server
    • D.Reviewing the provider's service-level agreement
    Show answerHide answer

    Correct answer: Consulting the cloud service provider's documentation

    Consulting the cloud service provider's documentation comes first here: the provider's guides list known issues, service limits and troubleshooting steps for that service. Rebooting the server that hosts the affected application changes the system before the problem is understood and can destroy the evidence. Checking the physical network connections to the server is not possible for the customer, because the provider owns and manages the physical layer. Reviewing the provider's service-level agreement tells you what performance the provider promises, but it does not help diagnose why the service is slow.

  4. In cloud-based virtual environments, which issue is most likely to cause a virtual machine (VM) to become unresponsive?

    • A.Overwhelmed host resources
    • B.Unsynchronized host clocks
    • C.Disabled memory ballooning
    • D.Disabled live migration
    Show answerHide answer

    Correct answer: Overwhelmed host resources

    Overwhelmed host resources are the likely cause: when the physical host's CPU, memory or storage I/O is saturated by competing VMs, it cannot schedule enough time for the guest, and the VM stops responding. Unsynchronized host clocks pass the wrong time to guests and cause authentication and logging errors (for example, Kerberos failures), not a hung VM. Disabled memory ballooning only removes one way for the hypervisor to reclaim memory; it does not freeze a running guest. Disabled live migration means the VM cannot be moved between hosts without downtime, but it does not affect how the VM runs.

  5. What is a common cause of performance degradation in a cloud storage service?

    • A.Expired IAM access keys
    • B.Exhausted EBS volume quotas
    • C.Enabled MFA delete
    • D.Exceeded IOPS limits
    Show answerHide answer

    Correct answer: Exceeded IOPS limits

    Exceeded IOPS limits are a common cause of slow storage: once a volume or tier reaches its maximum input/output operations per second, further requests are queued or throttled and latency rises. Expired IAM access keys produce authentication errors (access denied), not slower reads and writes. Exhausted EBS volume quotas stop new volumes from being created or enlarged; existing volumes keep their speed. Enabled MFA delete only requires a second factor before object versions can be permanently deleted, which has no effect on read or write performance.

  6. In a cloud environment, what is a common cause of an application failing to connect to its database?

    • A.Stale database statistics
    • B.Fragmented storage volumes
    • C.Oversized instance types
    • D.Restrictive firewall rules
    Show answerHide answer

    Correct answer: Restrictive firewall rules

    Restrictive firewall rules are the common cause: a rule that blocks the database port or the application's IP range stops the connection from being made at all. Stale database statistics lead the query planner to choose poor plans, so queries run slowly, but the connection still succeeds. Fragmented storage volumes slow disk access rather than preventing a connection. Oversized instance types waste money on unused capacity; they do not stop the application from reaching its database.

  7. What troubleshooting step is crucial when a cloud-based application experiences intermittent connectivity issues?

    • A.Increasing network bandwidth
    • B.Investigating network traffic
    • C.Restarting network services
    • D.Recreating network interfaces
    Show answerHide answer

    Correct answer: Investigating network traffic

    Investigating network traffic is the crucial step, because intermittent connectivity problems usually come from congestion, packet loss, retransmissions or a flapping route, and only traffic data shows when and where the drops happen. Increasing network bandwidth is a fix applied before the cause is known; if the drops come from a faulty route or a misconfigured rule, more bandwidth changes nothing. Restarting network services may clear the symptom for a while but does not find the cause, so the problem returns. Recreating network interfaces is a disruptive change that destroys the evidence and should only follow data showing the interface is at fault.

  8. Which action should be taken first when a cloud-based application's response time is slower than usual?

    • A.Doubling the network bandwidth
    • B.Upgrading the instance
    • C.Reviewing the application logs
    • D.Switching the hosting region
    Show answerHide answer

    Correct answer: Reviewing the application logs

    Reviewing the application logs comes first, because the application's logs show errors, slow queries, timeouts and resource warnings that point to the real cause before anything is changed. Doubling the network bandwidth spends money on a guess; if the bottleneck is in the code or the database, it changes nothing. Upgrading the instance is also a fix applied before the cause is known. Switching the hosting region is a major change that should only follow evidence that the region itself is the problem.

  9. What is a typical symptom of a misconfigured security group in a cloud environment?

    • A.Throttled disk throughput
    • B.Rising memory utilization
    • C.Blocked remote connections
    • D.Corrupted volume snapshots
    Show answerHide answer

    Correct answer: Blocked remote connections

    Blocked remote connections are the typical symptom: a security group acts as a virtual firewall, so a missing or wrong inbound rule stops SSH, RDP or application traffic from reaching the instance. Throttled disk throughput comes from storage IOPS or bandwidth limits, not from network rules. Rising memory utilization points to a workload or a memory leak inside the instance. Corrupted volume snapshots are a backup or storage problem; security group rules never touch snapshot data.

  10. In a cloud environment, what is the likely cause of a sudden increase in billing or resource usage costs?

    • A.Misconfigured auto-scaling rules
    • B.Misconfigured health check intervals
    • C.Misconfigured account lockout thresholds
    • D.Misconfigured load balancing weights
    Show answerHide answer

    Correct answer: Misconfigured auto-scaling rules

    Misconfigured auto-scaling rules are the likely cause: a scale-out threshold set too low, a missing scale-in rule or a maximum set too high keeps launching instances that are billed whether or not they are needed. Misconfigured health check intervals change how quickly failed instances are detected, which affects availability rather than the bill. Misconfigured account lockout thresholds lock users out after too few or too many failed logins, a security and access problem. Misconfigured load balancing weights send an uneven share of traffic to some servers, overloading them, but they do not add billable resources.

  11. What tool or method is most effective for identifying a memory leak in a cloud-based application?

    • A.Packet capture
    • B.Synthetic monitoring
    • C.Runtime profiling
    • D.Vulnerability scanning
    Show answerHide answer

    Correct answer: Runtime profiling

    Runtime profiling is the most effective method: a memory profiler records allocations and heap usage while the application runs, so it shows which objects keep growing and are never released. Packet capture records network traffic, which says nothing about memory inside the process. Synthetic monitoring runs scripted transactions to check availability and response time; it can show the slowdown a leak causes but not where the leak is. Vulnerability scanning looks for known security weaknesses, not memory faults.

  12. In cloud environments, which of the following is a common cause of DNS resolution failures?

    • A.Incorrect proxy server settings
    • B.Incorrect license server settings
    • C.Incorrect name server settings
    • D.Incorrect backup server settings
    Show answerHide answer

    Correct answer: Incorrect name server settings

    Incorrect name server settings are the common cause: if an instance or VPC points to the wrong DNS resolver, or to one it cannot reach, every lookup fails. Incorrect proxy server settings break outbound web requests that go through the proxy, but name lookups made by the host still work. Incorrect license server settings make licensed software fail its license check, not name resolution. Incorrect backup server settings cause backup jobs to fail or go to the wrong target, which does not affect DNS lookups.

  13. When experiencing performance issues in a cloud-based database, what is the first metric that should be checked?

    • A.Network I/O
    • B.CPU utilization
    • C.Memory utilization
    • D.Disk I/O
    Show answerHide answer

    Correct answer: Disk I/O

    Disk I/O is the metric to check first: databases constantly read and write data and logs, so slow or saturated storage (high latency, long queues, IOPS limits reached) is the most common bottleneck. Network I/O matters for client traffic and replication but is rarely the first limit a database hits. CPU utilization climbs with complex queries, but high CPU on a database is often a side effect of missing indexes, which also force heavy disk reads. Memory utilization is worth watching for cache efficiency, yet a database normally uses most of its memory by design, so a high reading alone is not a fault.

  14. Which component should be primarily investigated when a cloud-based virtual machine experiences slow boot times?

    • A.Security group rules
    • B.CPU allocation
    • C.Memory capacity
    • D.Storage performance
    Show answerHide answer

    Correct answer: Storage performance

    Storage performance should be investigated first, because booting means reading the operating system and its services from the boot volume; a slow, throttled or undersized disk makes every boot slow. Security group rules control network traffic once the VM is running and do not affect how fast it starts. CPU allocation sets how much processing power the VM has, but a boot spends most of its time waiting on reads from the boot volume, so more CPU does little to speed it up. Memory capacity decides how much the VM can cache and run at once; once there is enough to load the operating system, adding more does not shorten the boot.

  15. In cloud computing, what is a common cause of service interruption during a scale-down operation?

    • A.Overly long scaling cooldown periods
    • B.Premature instance terminations
    • C.Overly high minimum capacity
    • D.Enabled scale-in protection
    Show answerHide answer

    Correct answer: Premature instance terminations

    Premature instance terminations are the common cause: if instances are removed while they still handle critical work or open connections (for example, without connection draining or lifecycle hooks), users see dropped requests and outages. Overly long scaling cooldown periods make the group react slowly to demand, which wastes money or delays capacity but does not cut off running work. Overly high minimum capacity keeps more instances running than needed, which raises costs rather than interrupting service. Enabled scale-in protection stops the protected instances from being terminated during scale-in, so it prevents this kind of interruption.

  16. What is a likely cause of a sudden decrease in cloud application performance during peak usage times?

    • A.Ineffective scaling policies
    • B.Excessive memory allocation
    • C.Disabled connection draining
    • D.Oversized storage volumes
    Show answerHide answer

    Correct answer: Ineffective scaling policies

    Ineffective scaling policies are the likely cause: if the maximum capacity is too low, or the scale-out trigger reacts too late, the application cannot add instances fast enough when demand peaks, and performance drops. Excessive memory allocation gives the application more memory than it needs, which wastes money but does not slow it at peak. Disabled connection draining causes dropped requests when instances are removed or replaced, a problem tied to scale-in and deployments rather than to peak load. Oversized storage volumes cost more than necessary but do not slow the application under load.

  17. What is the primary reason for a "502 Bad Gateway" error in a cloud-based web application?

    • A.A misconfigured load balancer
    • B.A timed-out upstream response
    • C.A deleted web page
    • D.A revoked access token
    Show answerHide answer

    Correct answer: A misconfigured load balancer

    A misconfigured load balancer is the primary reason: a 502 Bad Gateway means a proxy or load balancer received an invalid response from the server behind it, which typically happens when its target groups, ports, protocols or health checks are set up wrong. A timed-out upstream response produces a 504 Gateway Timeout instead. A deleted web page leads to a 404 Not Found. A revoked access token leads to a 401 Unauthorized or 403 Forbidden response, not a gateway error.

  18. In cloud environments, which action should be taken to resolve throughput issues with a storage service?

    • A.Rotating the access keys
    • B.Upgrading the performance tier
    • C.Extending the retention period
    • D.Replicating the bucket objects
    Show answerHide answer

    Correct answer: Upgrading the performance tier

    Upgrading the performance tier resolves throughput problems, because cloud storage tiers (for example, standard versus provisioned or premium SSD) come with set throughput and IOPS ceilings, and a higher tier raises them. Rotating the access keys is a security practice that changes credentials, not speed. Extending the retention period keeps data or backups for longer, which affects cost and compliance rather than throughput. Replicating the bucket objects to another location improves durability and availability, but the original storage keeps the same throughput limits.

  19. What is a typical reason for HTTP request timeouts in a cloud application?

    • A.Overloaded or misconfigured web servers
    • B.Expired or revoked digital certificates
    • C.Missing or invalid credentials
    • D.Unregistered or expired domain names
    Show answerHide answer

    Correct answer: Overloaded or misconfigured web servers

    Overloaded or misconfigured web servers are the typical reason: when the web servers run out of threads, CPU or connections, or are set up with the wrong ports or limits, they cannot answer within the client's timeout. Expired or revoked digital certificates make the TLS handshake fail at once with a certificate error rather than a timeout. Missing or invalid credentials are rejected quickly with a 401 or 403 response. Unregistered or expired domain names fail at the DNS lookup, before any HTTP request is sent.

  20. Which troubleshooting step is critical when a cloud service is not accessible from certain geographical locations?

    • A.Reviewing the IAM permissions
    • B.Monitoring the CPU utilization
    • C.Patching the server OS
    • D.Checking the CDN configuration
    Show answerHide answer

    Correct answer: Checking the CDN configuration

    Checking the CDN configuration is the critical step, because a content delivery network serves users from regional edge locations; geo-restriction rules, a misconfigured edge location or regional routing errors can block access from some places while others work normally. Reviewing the IAM permissions matters when specific users are denied, not when whole regions are cut off. Monitoring the CPU utilization shows whether the server is busy, which would slow every user, not only those in certain locations. Patching the server OS fixes security and stability issues on the server but does not explain region-specific failures.

  21. What is the primary cause of SSL/TLS handshake failures in a cloud-based application?

    • A.Blocked or closed ports
    • B.Expired or invalid certificates
    • C.Weak or reused account passwords
    • D.Slow or overloaded database queries
    Show answerHide answer

    Correct answer: Expired or invalid certificates

    Expired or invalid certificates are the primary cause: during the handshake the client checks the server's certificate, and an expired certificate, a name mismatch or an untrusted issuer makes it abort the secure connection. Blocked or closed ports stop the connection before any handshake starts, so the client sees a timeout or a refused connection instead. Weak or reused account passwords are an account-security risk, not a handshake problem. Slow or overloaded database queries delay the application's responses after the secure session is already established.

  22. In a cloud environment, what could be the reason for a sudden spike in the error rate of API calls?

    • A.Expanded log storage
    • B.Increased CPU core count
    • C.Stale cached responses
    • D.Exceeded API rate limits
    Show answerHide answer

    Correct answer: Exceeded API rate limits

    Exceeded API rate limits are the likely reason: cloud APIs cap how many calls a client may make in a period, and once callers go over that cap the provider throttles the excess calls and returns errors such as HTTP 429 Too Many Requests, so the error rate jumps. Expanded log storage keeps more logs and raises storage costs but does not make API calls fail. An increased CPU core count gives the service extra processing capacity, which helps it keep up with requests rather than making them fail. Stale cached responses return out-of-date data with a success code, so they do not raise the error rate.

  23. In a cloud computing environment, what is a common cause of virtual machine (VM) migration failures?

    • A.Unpatched guest operating systems
    • B.Incompatible disk image formats
    • C.Missing cost allocation tags
    • D.Mismatched time zone settings
    Show answerHide answer

    Correct answer: Incompatible disk image formats

    Incompatible disk image formats are a common cause of failed migrations: if the source uses a format or hypervisor (for example, VMDK) that the target platform cannot import, the VM must be converted first or the migration fails. Unpatched guest operating systems are a security risk but migrate like any other VM. Missing cost allocation tags make spending harder to track, yet they do not stop a VM from moving. Mismatched time zone settings change the timestamps the guest shows, not whether it can be migrated.

DevOps Fundamentals (19)

  1. In cloud computing, what is the primary purpose of orchestration?

    • A.To convert repetitive processes into reusable scripts
    • B.To combine automated tasks into coordinated workflows
    • C.To transform resource metrics into performance alerts
    • D.To process usage records into departmental invoices
    Show answerHide answer

    Correct answer: To combine automated tasks into coordinated workflows

    To combine automated tasks into coordinated workflows is the purpose of orchestration: it arranges many automated jobs across systems, middleware and services so they run in the right order as one managed process. To convert repetitive processes into reusable scripts is automation, which handles individual tasks; orchestration sits a level above it and strings those automated tasks together. To transform resource metrics into performance alerts is the job of monitoring tools. To process usage records into departmental invoices is billing and chargeback work, not orchestration.

  2. In cloud computing, what is the primary function of container orchestration?

    • A.Building and publishing container images to registries
    • B.Deploying and managing containers across cluster nodes
    • C.Scanning and patching container images for flaws
    • D.Packaging and isolating applications as containers
    Show answerHide answer

    Correct answer: Deploying and managing containers across cluster nodes

    Deploying and managing containers across cluster nodes is what container orchestration does: a tool such as Kubernetes schedules containers onto nodes and handles their whole lifecycle, including scaling, networking, restarts and load balancing. Building and publishing container images to registries is the work of a build pipeline and an image registry. Scanning and patching container images for flaws is container security scanning. Packaging and isolating applications as containers is containerization itself, done by a container runtime such as Docker, which orchestration then manages at scale.

  3. In cloud environments, what is the primary purpose of orchestration tools?

    • A.To watch servers and alert staff when they go down
    • B.To encrypt data held on storage volumes
    • C.To record which users opened each file
    • D.To automate deployment of resources
    Show answerHide answer

    Correct answer: To automate deployment of resources

    Orchestration tools exist to automate deployment of resources, coordinating many provisioning and configuration steps into one repeatable workflow. 'To watch servers and alert staff when they go down' is the job of a monitoring tool; 'to encrypt data held on storage volumes' is storage encryption; 'to record which users opened each file' is audit logging. None of these coordinates deployment.

  4. Which tool in cloud management is primarily used for automating application deployment, scaling, and operations?

    • A.Terraform
    • B.Jenkins
    • C.CloudFormation
    • D.Kubernetes
    Show answerHide answer

    Correct answer: Kubernetes

    Kubernetes is the open-source platform that automates deploying, scaling and operating containerized applications across a cluster. Terraform provisions infrastructure from code but does not run or scale the applications; Jenkins runs CI/CD pipelines that build, test and ship code but does not scale or keep applications running; CloudFormation is AWS's template service for provisioning infrastructure. Each is a real automation or operations tool, but only one orchestrates application containers.

  5. In a CI/CD pipeline, what does the 'continuous integration' (CI) stage primarily accomplish?

    • A.Releasing passing commits into production automatically and skipping approvals and sign-offs
    • B.Bundling every tested release into a deployable package and awaiting manual review and approval
    • C.Streaming metrics from live services into central dashboards and alerting on errors and outages
    • D.Merging code changes into a shared repository frequently and running automated builds and tests
    Show answerHide answer

    Correct answer: Merging code changes into a shared repository frequently and running automated builds and tests

    Continuous integration means 'Merging code changes into a shared repository frequently and running automated builds and tests', so integration problems surface within minutes of each merge and the main branch stays stable. 'Releasing passing commits into production automatically and skipping approvals and sign-offs' describes continuous deployment, a later stage of the pipeline. 'Bundling every tested release into a deployable package and awaiting manual review and approval' describes continuous delivery. 'Streaming metrics from live services into central dashboards and alerting on errors and outages' is continuous monitoring of production, which happens after release rather than at integration.

  6. What is the key difference between continuous delivery and continuous deployment in a CI/CD pipeline?

    • A.Continuous deployment waits for a person to approve each production release, while continuous delivery releases automatically
    • B.Continuous delivery runs automated tests against each production release, while continuous deployment skips the testing stage
    • C.Continuous delivery waits for a person to approve each production release, while continuous deployment releases automatically
    • D.Continuous deployment runs automated tests against each production release, while continuous delivery skips the testing stage
    Show answerHide answer

    Correct answer: Continuous delivery waits for a person to approve each production release, while continuous deployment releases automatically

    'Continuous delivery waits for a person to approve each production release, while continuous deployment releases automatically' is the defining difference: in continuous delivery every change is built, tested and made ready to ship, but a human triggers the production release, whereas continuous deployment pushes every change that passes the pipeline straight to production. 'Continuous deployment waits for a person to approve each production release, while continuous delivery releases automatically' reverses the two terms. 'Continuous delivery runs automated tests against each production release, while continuous deployment skips the testing stage' and 'Continuous deployment runs automated tests against each production release, while continuous delivery skips the testing stage' are both wrong because both practices rely on the same automated test stages; continuous deployment in particular depends on thorough automated testing, since no person reviews the release.

  7. A developer creates a branch, makes changes, and opens a request for teammates to review and approve before merging into the main branch. What is this version control mechanism called?

    • A.A merge commit
    • B.A pull request
    • C.A change request
    • D.A revert commit
    Show answerHide answer

    Correct answer: A pull request

    'A pull request' (called a merge request in GitLab) proposes merging a branch into another and gives teammates a place to review, discuss and approve the change, often with automated checks, before the merge happens. 'A merge commit' is the commit Git records after branches are combined, so it comes after the review rather than requesting it. 'A change request' is the IT service management form submitted for approval of a change to production systems, often by a change advisory board, not the version control mechanism for reviewing a branch. 'A revert commit' is a new commit that undoes an earlier one and involves no review before merging.

  8. Which characteristic best distinguishes a container from a traditional virtual machine?

    • A.Containers share the host operating system kernel and package an application with its dependencies
    • B.Containers boot a complete guest operating system under a hypervisor and emulate physical hardware
    • C.Containers each run their own kernel for stronger isolation on top of the host operating system
    • D.Containers start quickly because they resume a paused snapshot of a full guest operating system
    Show answerHide answer

    Correct answer: Containers share the host operating system kernel and package an application with its dependencies

    'Containers share the host operating system kernel and package an application with its dependencies', which makes them lightweight and quick to start compared with virtual machines. 'Containers boot a complete guest operating system under a hypervisor and emulate physical hardware' describes a virtual machine, which is exactly what a container avoids. 'Containers each run their own kernel for stronger isolation on top of the host operating system' is wrong because all containers on a host use the one host kernel, which is why their isolation is generally weaker than a VM's. 'Containers start quickly because they resume a paused snapshot of a full guest operating system' is wrong because a container starts fast simply because there is no guest operating system to boot.

  9. Which tool is an open-source platform for automating the deployment, scaling, and management of containerized applications?

    • A.Ansible
    • B.Jenkins
    • C.Prometheus
    • D.Kubernetes
    Show answerHide answer

    Correct answer: Kubernetes

    Kubernetes is the open-source container orchestration platform that automates deploying, scaling, self-healing and managing containerized applications across a cluster of hosts. Ansible automates configuration and can start containers from a playbook, but it does not continuously schedule, scale and heal them across a cluster. Jenkins is an automation server for CI/CD pipelines, and Prometheus collects and stores metrics for monitoring and alerting; neither one orchestrates containers.

  10. Which DevOps tool is most commonly associated with automating CI/CD pipelines and build/test/deploy jobs?

    • A.Git
    • B.Ansible
    • C.Jenkins
    • D.Grafana
    Show answerHide answer

    Correct answer: Jenkins

    Jenkins is the open-source automation server most closely associated with CI/CD; its pipelines run the build, test and deploy jobs that move code from commit to release. Git is the version control system whose commits trigger those pipelines, but it does not run the jobs. Ansible automates configuration management and can be called from a pipeline step, yet it is not a pipeline server. Grafana builds dashboards from metrics and plays no part in building or testing code.

  11. A team uses a configuration management tool that applies an agentless, declarative approach using YAML playbooks to configure servers. Which DevOps tool fits this description?

    • A.Puppet
    • B.Ansible
    • C.Chef
    • D.Terraform
    Show answerHide answer

    Correct answer: Ansible

    Ansible is the agentless configuration management tool that describes the desired state of servers in YAML playbooks and applies them over SSH, so nothing has to be installed on the managed nodes. Puppet is also declarative but normally relies on an agent on each node and uses its own Puppet language rather than YAML playbooks. Chef uses Ruby-based recipes and cookbooks with a client agent on each node. Terraform is declarative and agentless, but it provisions infrastructure from HCL files rather than configuring servers with playbooks.

  12. In Git-based version control, what is the primary purpose of branching?

    • A.To mark a fixed point in the commit history that identifies a release version
    • B.To shelve uncommitted edits on a temporary stack that keeps them for reuse
    • C.To copy an entire repository into a separate user account that tracks the original
    • D.To let work proceed on a parallel line of development that is merged when ready
    Show answerHide answer

    Correct answer: To let work proceed on a parallel line of development that is merged when ready

    Branching exists 'To let work proceed on a parallel line of development that is merged when ready': a feature or fix is developed on its own branch without disturbing the main branch, then merged back once it is ready. 'To mark a fixed point in the commit history that identifies a release version' describes a tag. 'To shelve uncommitted edits on a temporary stack that keeps them for reuse' describes git stash. 'To copy an entire repository into a separate user account that tracks the original' describes a fork, which copies the whole repository rather than opening a line of work inside it.

  13. A CI/CD pipeline produces a packaged, versioned build output that is stored for later deployment. What is this output called?

    • A.Manifest
    • B.Release tag
    • C.Build agent
    • D.Artifact
    Show answerHide answer

    Correct answer: Artifact

    An artifact is the packaged, versioned output of a build stage, such as a compiled binary, an archive or a container image, kept in an artifact repository so later pipeline stages can deploy exactly the same build. A 'Manifest' is a file that lists or declares the contents and settings of a package or deployment; it describes the output rather than being the deployable package itself. A 'Release tag' is a label on a commit in version control that marks a version of the source code. A 'Build agent' is the machine or container that runs the build jobs, not what they produce.

  14. Which open-source tool is most commonly used for visualizing metrics and building observability dashboards in a DevOps environment?

    • A.Grafana
    • B.Prometheus
    • C.Nagios
    • D.Splunk
    Show answerHide answer

    Correct answer: Grafana

    Grafana is the open-source tool most widely used to visualize metrics and build observability dashboards, drawing on data sources such as Prometheus, Loki or cloud monitoring services. Prometheus collects and stores time-series metrics and evaluates alert rules; it is usually the data source behind the dashboards rather than the dashboard tool. Nagios checks hosts and services and raises alerts, with only basic status views. Splunk does offer dashboards, but it is a commercial log search and analytics platform rather than the open-source dashboard tool the question describes.

  15. What is the main goal of incorporating automated security scanning (DevSecOps) into a CI/CD pipeline?

    • A.To shift vulnerability testing earlier and catch flaws automatically before code ships
    • B.To hold the build for a manual sign-off by auditors and ship it once other checks pass
    • C.To monitor running workloads and prevent intrusions through a web application firewall
    • D.To encrypt artifacts at rest and restrict pipeline access to approved administrators
    Show answerHide answer

    Correct answer: To shift vulnerability testing earlier and catch flaws automatically before code ships

    DevSecOps adds automated security scans (static analysis, dependency and image scanning) to the pipeline itself, so its goal is 'To shift vulnerability testing earlier and catch flaws automatically before code ships', the shift-left approach. 'To hold the build for a manual sign-off by auditors and ship it once other checks pass' keeps security as a manual gate at the end, which is the practice DevSecOps replaces. 'To monitor running workloads and prevent intrusions through a web application firewall' is runtime protection after release, not scanning in the pipeline. 'To encrypt artifacts at rest and restrict pipeline access to approved administrators' hardens the pipeline itself but does not find vulnerabilities in the code being built.

  16. A cloud engineering team stores its Terraform configuration files in a shared repository so that every change is recorded, can be attributed to an author, and can be rolled back to a previous state if a deployment breaks. Which type of tool provides this capability?

    • A.A build server such as Jenkins
    • B.A package registry such as Artifactory
    • C.A version control system such as Git
    • D.A configuration management tool such as Ansible
    Show answerHide answer

    Correct answer: A version control system such as Git

    A version control system such as Git records every change to the files, who made it and when, and lets the team review history and revert to an earlier commit if a deployment breaks; it is the standard home for infrastructure-as-code. A build server such as Jenkins runs pipelines when code changes but does not keep the files' history. A package registry such as Artifactory stores versioned build outputs, not the authored history of configuration files. A configuration management tool such as Ansible applies settings to servers rather than tracking changes to the Terraform code.

  17. In a CI/CD pipeline, a developer's commit automatically triggers a build that compiles the code and runs the unit-test suite, merging the change into the mainline only if every test passes. Which part of CI/CD does this describe?

    • A.Continuous deployment
    • B.Continuous integration
    • C.Continuous delivery
    • D.Continuous improvement
    Show answerHide answer

    Correct answer: Continuous integration

    Continuous integration is described: each commit triggers an automated build and unit tests, and the change joins the shared mainline only when every test passes. Continuous delivery and continuous deployment are the later CD stages that package the tested build and release it toward staging or production, which this scenario never reaches. Continuous improvement is the ongoing practice of refining processes over time, not a pipeline stage.

  18. A cloud architect needs to run many isolated instances of a microservice on a single Linux host while keeping startup time in milliseconds and memory overhead minimal. Compared with running each instance in its own virtual machine, why does packaging each microservice as a container achieve this goal?

    • A.Containers run their own guest OS, while each VM shares the host kernel and packages only the app and its dependencies
    • B.Containers run on a hypervisor that emulates the host OS kernel and drivers, while each VM drops isolation to save RAM
    • C.Containers share the host kernel and package only the app and its dependencies, while each VM carries its own guest OS
    • D.Containers drop isolation to save RAM, while each VM runs on a hypervisor that emulates the host OS kernel and drivers
    Show answerHide answer

    Correct answer: Containers share the host kernel and package only the app and its dependencies, while each VM carries its own guest OS

    'Containers share the host kernel and package only the app and its dependencies, while each VM carries its own guest OS' explains the gain: with no guest operating system to boot per instance, containers start in milliseconds and use far less memory, so many fit on one host. 'Containers run their own guest OS, while each VM shares the host kernel and packages only the app and its dependencies' reverses the two. 'Containers run on a hypervisor that emulates the host OS kernel and drivers, while each VM drops isolation to save RAM' is wrong on both counts: containers need no hypervisor, and VMs keep strong isolation from each other. 'Containers drop isolation to save RAM, while each VM runs on a hypervisor that emulates the host OS kernel and drivers' is also wrong: containers stay isolated through namespaces and cgroups, and a hypervisor presents virtual hardware on which each VM boots its own guest kernel rather than emulating the host kernel.

  19. An organization runs production workloads across AWS, Azure, and Google Cloud and wants a single infrastructure-as-code tool and language to provision resources consistently in all three. Which choice best meets this requirement, and why?

    • A.AWS CloudFormation, because it tracks provisioned stacks internally with no separate state file
    • B.HashiCorp Terraform, because it uses one declarative language with providers for many platforms
    • C.Azure Bicep, because it offers a concise syntax with no separate state file
    • D.AWS CDK, because it lets engineers define stacks with general-purpose programming languages
    Show answerHide answer

    Correct answer: HashiCorp Terraform, because it uses one declarative language with providers for many platforms

    HashiCorp Terraform, because it uses one declarative language with providers for many platforms, is the right choice: the same HashiCorp Configuration Language (HCL) code and workflow provision resources in AWS, Azure and Google Cloud through each cloud's provider. AWS CloudFormation, because it tracks provisioned stacks internally with no separate state file, is true of CloudFormation, but it provisions AWS resources only. Azure Bicep, because it offers a concise syntax with no separate state file, is true of Bicep, but it deploys only to Azure. AWS CDK, because it lets engineers define stacks with general-purpose programming languages, is true of the CDK, but it deploys through CloudFormation and so stays inside AWS.

References

  1. 1.CompTIA. “Cloud+ Certification (official; CV0-004 domains and weights).” CompTIA, 2026. ↑
  2. 2.CompTIA. “Cloud+ CV0-004 Exam Objectives (official PDF).” CompTIA. ↑
  3. 3.CompTIA. “The New CompTIA Cloud+ (CV0-004 changes).” CompTIA. ↑
  4. 4.CompTIA. “How to Study for CompTIA Cloud+.” CompTIA. ↑
  5. 5.CompTIA. “Marketplace - Exam Vouchers and Pricing.” CompTIA. ↑
Career Employer

Career Employer is the ultimate resource to help you get started working the job of your dreams. We cover topics from general career information, career searching, exam preparation with free study materials, career interviewing, and becoming successful in your career of choice.

Follow Us:

All Posts

Career Employer’s Editorial Process

Here at Career Employer, we focus a lot on providing factually accurate information that is always up to date. We strive to provide correct information using strict editorial processes, article editing, and fact-checking for all of the information found on our website. We only utilize trustworthy and relevant resources. To find out more, make sure to read our full editorial process page here.