A team creates an Amazon S3 bucket in the us-east-1 Region and assumes that AWS will automatically keep a copy of the objects in the eu-west-1 Region for resilience. Which statement is correct?
Intermediate · Choose one
Explanation
Answer: B. AWS Regions are designed to be isolated from each other so that a problem in one Region does not affect another, and so that customers retain control over where their data resides. AWS does not move or copy customer content between Regions unless the customer configures a feature that does so, such as S3 Cross-Region Replication or a snapshot copy.
A. No automatic cross-Region copying occurs, and cross-Region replication incurs both storage and data transfer charges when it is enabled.
C. A bucket is created in one Region and its data resides there. Although bucket names are globally unique, the bucket itself is a Regional resource.
D. There is no continent-level automatic replication. Multi-Region resilience is always an explicit customer configuration.
22
An application has a web tier and a database tier that exchange a high volume of chatty, latency-sensitive traffic. The business requires the application to survive the loss of an entire data center, but it wants to avoid cross-Region data transfer charges and added network latency between the tiers. What should the architect do?
Intermediate · Choose one
Explanation
Answer: A. Availability Zones are physically separate facilities with independent power, cooling, and networking, so spreading the tiers across two or more Availability Zones survives the loss of an entire data center. Because Availability Zones in a Region are linked by high-bandwidth, low-latency private networking, inter-tier chatter stays fast, and no cross-Region data transfer charges are incurred.
B. Splitting tiers across Regions puts hundreds or thousands of miles between the web and database tiers, which adds latency to every query and incurs cross-Region data transfer charges. Multi-Region designs replicate whole stacks rather than splitting a single stack.
C. Edge locations serve cached content and do not host an application's web tier, and leaving the database in one Availability Zone still leaves a single point of failure.
D. A single Availability Zone is a single data center location, so the design fails the requirement. Snapshots help recovery but do not keep the application running.
23
A hospital must keep certain patient processing workloads physically inside its own building to satisfy regulators, but it wants to use the same AWS services, APIs, and management tools it uses in the cloud. Which AWS offering meets this requirement?
Intermediate · Choose one
Explanation
Answer: C. AWS Outposts delivers AWS-designed and AWS-managed infrastructure racks or servers to the customer's own data center or colocation space. It runs a consistent set of AWS services locally and uses the same APIs, control plane, and tools as the AWS Region, which makes it the fit for data that must physically remain on premises.
A. Storage Gateway provides on-premises access to cloud storage, but it addresses only storage, not the compute workload the hospital must run locally.
B. Amazon Lightsail is a simplified way to run virtual servers within an AWS Region. It has no on-premises deployment option.
D. AWS Direct Connect provides a dedicated private network link from the hospital to an AWS Region. It moves traffic to AWS rather than allowing the workload itself to run inside the hospital building.
24
Which TWO of the following are valid reasons for a company to deploy the same workload in more than one AWS Region? (Choose TWO.)
Intermediate · Select 2 answers
Explanation
Answer: D and E. Regions are isolated from one another, so running in more than one Region is the standard approach to surviving a Region-wide event as part of a disaster recovery strategy. Multiple Regions also let a company place data in a specific country to satisfy data sovereignty or residency laws, and they can reduce latency for users spread across the world.
A. Amazon S3 pricing varies by Region and is not discounted simply because a customer uses more than one Region.
B. Replication across Regions protects against a Region failure, but it also replicates accidental deletions and corruption. Backups remain necessary.
C. IAM is a global service, but permissions and resource policies still have to be designed and applied for the resources in every Region. Nothing is removed.
25
A gaming company runs UDP-based multiplayer game servers on Amazon EC2 in three AWS Regions. It needs two static IP addresses that never change, wants player traffic to enter the AWS global network as close to the player as possible, and needs near-instant failover if a Regional endpoint becomes unhealthy. Which service should it use?
Intermediate · Choose one
Explanation
Answer: B. AWS Global Accelerator provides two static anycast IP addresses that act as a fixed front door for the application. Traffic enters the AWS global network at the edge location nearest the user and is routed over the AWS backbone to the healthiest endpoint. It supports TCP and UDP, and because clients keep using the same static IPs, failover between Regions does not depend on DNS caching.
A. An Application Load Balancer is Regional and operates at Layer 7 for HTTP and HTTPS. It neither handles UDP nor provides a global entry point.
C. Amazon CloudFront is optimized for cacheable HTTP and HTTPS content. It does not front UDP game traffic.
D. Route 53 is DNS-based. Failover depends on TTL expiry and client-side DNS caching, so it is not near-instant, and it does not provide static anycast IP addresses.
26
A company is migrating a large in-memory analytics database that requires a very high ratio of RAM to vCPU. Which Amazon EC2 instance family is the best starting point?
Intermediate · Choose one
Explanation
Answer: C. Memory optimized instances (for example the R and X families) are designed to deliver fast performance for workloads that process large data sets in memory, such as high-performance relational and in-memory databases, real-time big data analytics, and in-memory caches.
A. Storage optimized instances (I, D, H families) provide high sequential read and write access to very large local data sets and low-latency local NVMe storage, not large amounts of RAM.
B. Accelerated computing instances (P, G, Inf, Trn families) use hardware accelerators such as GPUs for machine learning, graphics, and specialized parallel workloads.
D. Compute optimized instances (C family) offer a high ratio of vCPU to memory and suit batch processing, media transcoding, high-performance web servers, and scientific modeling.
27
A visual effects studio runs nightly rendering jobs that can be stopped and restarted without harm, and it wants the lowest possible Amazon EC2 compute cost. Which purchasing option best fits?
Intermediate · Choose one
Explanation
Answer: D. Spot Instances let you use spare EC2 capacity at discounts of up to 90 percent compared with On-Demand prices. AWS can reclaim the capacity with a two-minute interruption notice, so Spot is intended precisely for fault-tolerant, flexible, interruption-tolerant workloads such as batch rendering, CI builds, and data analysis.
A. Dedicated Hosts are the most expensive option and exist to satisfy licensing or compliance needs for physical server isolation, not to minimize cost.
B. A three-year Reserved Instance offers a deep discount, but it commits the company for three years and is intended for steady-state, always-on workloads rather than intermittent nightly jobs.
C. On-Demand carries no commitment but is the highest per-hour price of the flexible options, so it does not meet the lowest-cost goal.
28
A company has a steady baseline of compute usage but expects to change instance families and move workloads between Regions over the next year. It wants a discount in exchange for a one-year usage commitment while keeping that flexibility. Which purchasing option should it choose?
Intermediate · Choose one
Explanation
Answer: D. Compute Savings Plans commit the customer to a consistent amount of compute usage measured in dollars per hour for a one-year or three-year term, in exchange for savings of up to 66 percent for Compute Savings Plans and up to 72 percent for EC2 Instance Savings Plans. Compute Savings Plans automatically apply regardless of instance family, size, Availability Zone, Region, operating system, or tenancy, and also cover Fargate and Lambda usage.
A. Standard Reserved Instances give the deepest RI discount but lock the customer to a specific instance family and Region, which conflicts with the stated need for flexibility.
B. On-Demand Capacity Reservations reserve capacity in a specific Availability Zone but provide no billing discount on their own.
C. Spot Instances offer no commitment and no guaranteed availability. They can be reclaimed at any time and are unsuitable for a steady baseline.
29
A development team already runs Kubernetes on-premises and wants to move to AWS while continuing to use standard Kubernetes tooling and manifests, without operating the Kubernetes control plane themselves. Which service should they use?
Intermediate · Choose one
Explanation
Answer: D. Amazon EKS is a managed service that runs upstream, certified conformant Kubernetes. AWS operates the control plane across multiple Availability Zones, while the team keeps using standard Kubernetes APIs, manifests, and ecosystem tooling such as kubectl and Helm.
A. Amazon ECS is AWS's own container orchestrator. It is simpler for teams starting fresh, but it does not use Kubernetes APIs, so existing manifests and tooling would not carry over.
B. Elastic Beanstalk deploys applications onto managed EC2 environments. It is not a Kubernetes platform.
C. EC2 Auto Scaling manages the number of EC2 instances. It provides no container orchestration.
30
A team wants to run containerized microservices on AWS without provisioning, patching, or right-sizing any Amazon EC2 instances. Which service provides this?
Intermediate · Choose one
Explanation
Answer: C. AWS Fargate is a serverless compute engine for containers that works with both Amazon ECS and Amazon EKS. You specify the CPU and memory each task or pod needs, and AWS provisions and manages the underlying compute, so there are no EC2 instances to launch, patch, or scale.
A. Elastic Beanstalk provisions EC2 instances into the customer's account, so the customer still owns the servers.
B. Amazon ECR stores container images. It does not run them.
D. Using the ECS-optimized AMI means the customer is explicitly running and managing EC2 container instances, which is the opposite of the requirement.