Question 360
A batch-rendering farm can retry individual jobs and checkpoint progress to S3. Jobs have a flexible completion window, and on-demand EC2 charges dominate costs. Which purchase model is most suitable for incremental capacity?
- Use EC2 Spot Instances for interruption-tolerant workers with checkpointing, retry and capacity diversification.
- Use a three-year Compute Savings Plan sized to the one-week peak, assuming it carries no commitment cost during idle weeks.
- Use EC2 On-Demand capacity for all interruption-tolerant jobs solely to eliminate any need for checkpointing, despite checkpoints already existing.
- Use Spot only in one instance type and Availability Zone, treating the cheapest pool as a guarantee of availability.
Explanation
Correct Answer
A:Spot discounts match flexible stateless or checkpointed batch work, provided interruption handling is implemented.
Incorrect Answers
B:A Savings Plan is a continuing hourly dollar commitment; it can be wasteful for highly intermittent, short-lived peaks.
C:On-Demand avoids Spot interruptions but is generally more expensive for these retryable flexible jobs; the workload already supports safe Spot use.
D:Spot capacity varies and can be interrupted; diversification and retry/checkpoint handling reduce risks under this flexible batch requirement.
Question 361
An analytics application copies large objects between EC2 instances in different Availability Zones every hour. Cost and Usage Report data identifies substantial cross-AZ transfer charges. The architecture allows processing near the data producer without reducing fault tolerance of essential data storage. What action should be evaluated first?
- Enable VPC gateway endpoints for S3 and treat the EC2-to-EC2 inter-AZ transfer flow as S3 gateway endpoint traffic.
- Move all compute into one AZ and remove durable cross-AZ storage replicas, without assessing the fault-tolerance requirements.
- Increase enhanced networking bandwidth for the EC2 fleet and assume higher throughput decreases billed cross-AZ byte volume.
- Co-locate high-volume processing with its data source where safe, and measure cross-AZ bytes after preserving necessary multi-AZ resilience.
Explanation
Correct Answer
D:Localizing avoidable bulk transfers can reduce inter-AZ cost while maintaining redundancy where the workload needs it.
Incorrect Answers
A:An S3 gateway endpoint does not change the measured cross-AZ network path between EC2 instances; the charge driver is not S3 API egress.
B:Concentration may reduce bytes but can violate resilience. Co-locate only safe high-volume processing while retaining required multi-AZ data protection.
C:Interface throughput capability does not reduce the number of bytes crossing AZ boundaries, which drives the identified transfer charges.
Question 362
A newly launched DynamoDB application has unpredictable, bursty request volume and no reliable steady-state estimate. Engineers need to avoid capacity planning during initial adoption while still paying for actual read and write requests. Which billing mode should they consider?
- Choose DynamoDB on-demand capacity for unpredictable traffic and reassess against provisioned capacity when patterns stabilize.
- Use provisioned capacity configured for the highest imaginable peak for every hour until the new application stabilizes.
- Use provisioned capacity at the expected minimum without auto scaling and assume unexpected bursts will never be throttled.
- Enable provisioned capacity with auto scaling and assume it instantaneously handles every unforeseen one-second demand spike without warm-up or limits.
Explanation
Correct Answer
A:On-demand capacity avoids upfront RCU/WCU sizing and bills supported request usage, fitting variable demand during launch.
Incorrect Answers
B:Always-on peak provisioned units can create waste for a workload explicitly lacking predictable baseline demand.
C:Insufficient fixed provisioning can throttle during the observed unpredictable spikes, violating the low-planning adoption requirement.
D:Auto scaling can be appropriate for measured trends but is not a guarantee for abrupt unknown bursts and still introduces capacity planning/tuning.
Question 363
A team already delivers globally through CloudFront, but its cache hit ratio is unexpectedly low because each download request includes a random tracking query string. The files are immutable and the tracking parameter never changes their bytes. What cache configuration reduces redundant origin fetches?
- Include every tracking query-string value in the CloudFront cache key but shorten TTL to a few seconds.
- Forward the tracking parameter to the origin in an origin request policy while also requiring it as a distinct cache-key dimension.
- Change the CloudFront cache policy to exclude the irrelevant tracking query parameter from the cache key while retaining required authorization and content-varying inputs.
- Remove all query parameters from the cache key, including parameters that actually change content or authorization.
Explanation
Correct Answer
C:Reducing unnecessary cache-key variation lets identical immutable downloads reuse an edge object, while preserving fields that genuinely alter or authorize the response.
Incorrect Answers
A:The irrelevant random parameter still fragments the cache into variants; a shorter TTL can increase origin traffic further.
B:Forwarding for analytics can be separate from cache-key selection. Keeping irrelevant values in the cache key still reduces hit ratio.
D:Overly broad exclusion can serve an incorrect representation or cross a security boundary; remove only inputs that do not affect the response.
Question 364
A video rendering team already uses diversified EC2 Spot Instances with retry and S3 checkpoints. It plans a one-week, interruption-tolerant capacity expansion across several eligible Regions but cannot reliably predict where a requested mix of four instance families is most likely to be fulfilled. Before requesting the capacity, which first-party planning signal should the architect obtain?
- Calculate EC2 Spot placement scores for the target capacity and multiple eligible instance types/Regions, using the results as non-guaranteed likelihood recommendations.
- Rely only on the current Spot hourly price in one AZ and assume the cheapest AZ guarantees enough capacity for the entire upcoming week.
- Buy an EC2 On-Demand Capacity Reservation in one AZ and assume its guaranteed On-Demand placement is an unbiased Spot placement score for every Region.
- Use last week’s Spot interruption percentages alone as a guarantee of capacity availability for the requested multi-family mix.
Explanation
Correct Answer
A:Spot placement scores estimate likelihood of fulfilling an eligible target capacity across candidate Regions or AZs; the score is advisory, not a guarantee.
Incorrect Answers
B:Low current Spot price does not prove capacity availability for the specified multi-family request over the requested timeframe.
C:An On-Demand Capacity Reservation is a distinct capacity purchase, not an assessment of future Spot fulfillment likelihood.
D:Historic interruption frequency measures a different risk from whether a requested target capacity can be placed across the chosen Regions and types.
Question 365
A mature numerical-modeling service runs EC2 m7i instances exclusively in us-east-1 and has a stable dollar-per-hour requirement for three years. It may change between m7i sizes or Linux and Windows, but will not switch families or Regions, and does not use Lambda or Fargate. Which plan can provide the more specific family-and-Region discount commitment?
- An EC2 Instance Savings Plan committed to the eligible m7i family in us-east-1, sized to measured baseline usage.
- Choose a Compute Savings Plan for its broad Region and workload flexibility even though family and Region are fixed and the goal is the more specific commitment.
- Buy a three-year Standard Reserved Instance for one exact Linux instance configuration and assume all later Windows and size changes inherit its discount automatically.
- Create an EC2 On-Demand Capacity Reservation and count it as the instance-family Savings Plans discount for each billed hour.
Explanation
Correct Answer
A:EC2 Instance Savings Plans commit to a specified instance family and Region while applying across supported sizes, operating systems and tenancy.
Incorrect Answers
B:Compute Savings Plans can provide valid discounts but are the more flexible, typically less targeted choice; the question asks for the family-and-Region commitment with greater specific discounts.
C:A Standard RI is less flexible about platform and purchase attributes; it does not provide the same family/Region spend commitment across those OS changes.
D:Capacity Reservations reserve placement, while Savings Plans provide a separate spending discount; the former alone does not create the latter.
Question 366
A company stores monthly compliance evidence in S3. It is downloaded during the first week and almost never requested again, but must be retained for seven years; a restore time of several hours is acceptable for old evidence. What design reduces long-term storage spend?
- Transition every object to Deep Archive the moment it is written, before the known first-week evidence downloads finish.
- Use lifecycle rules to transition old objects to suitable archive storage classes while enforcing retention and restore requirements.
- Keep every document in S3 Standard for seven years without evaluating lower-cost classes despite extremely infrequent access.
- Set S3 expiration to seven months after creation because archive classes automatically satisfy a seven-year regulatory retention window.
Explanation
Correct Answer
B:Age-driven transitions match lower retrieval frequency with cheaper archival storage and preserve explicit required retention periods.
Incorrect Answers
A:Deep Archive requires restore before normal reads, which impairs the defined immediate-access period even if suitable after data becomes cold.
C:Standard is designed for frequent access and may cost more than suitable archive tiers for years of cold data.
D:A shorter expiration date deletes required evidence; changing the storage class cannot waive retention length.
Question 367
A compliance pipeline crawls a bucket with 70 million S3 objects every night using repeated synchronous ListObjectsV2 calls, though its report only needs a daily catalog of keys, sizes and storage classes. The pipeline can accept yesterday’s inventory and does not need a live list. What architecture should reduce API listing work?
- Configure daily S3 Inventory reports in a supported output format and run the catalog analysis from that delivered listing instead of repeatedly enumerating the full bucket.
- Deploy S3 Storage Lens and rely only on bucket-level storage aggregates instead of obtaining an object-key-and-size catalog.
- Use S3 Batch Operations without any existing object manifest and assume the batch API automatically creates a daily report of every key.
- Parallelize more ListObjectsV2 requests over the full 70-million-key namespace every night rather than use a delivered inventory.
Explanation
Correct Answer
A:S3 Inventory creates a scheduled object and metadata listing, which can replace unnecessary high-volume synchronous List API enumeration when daily freshness is sufficient.
Incorrect Answers
B:Storage Lens supports storage usage and activity metrics but is not a replacement for the required per-object inventory listing.
C:Batch Operations executes actions against an eligible manifest; it does not independently supply the requested daily metadata inventory.
D:Parallel requests may improve elapsed time but retain high listing volume and avoid none of the unnecessary synchronous enumeration.
Question 368
A private Fargate service pulls company-owned Linux images from Amazon ECR in the same Region and has no public base-image pull-through or foreign layers. Its image downloads traverse a NAT gateway; flow and cost reports show substantial NAT data-processing charges, while internet egress is prohibited by policy. What eligible VPC path should the architect assess for ECR API, registry and image layers?
- Use ECR interface endpoints for ecr.api and ecr.dkr with private DNS, required endpoint security groups and an S3 gateway endpoint for layer retrieval; compare endpoint costs with NAT.
- Create only an S3 gateway endpoint and assume it provides ECR authorization tokens and all Docker registry API calls without any ECR endpoints.
- Create only an ECR API interface endpoint and assume it covers Docker registry requests plus every S3 layer download by itself.
- Place NAT gateways in additional Availability Zones to serve the private ECR image pull and assume public egress is now prohibited.
Explanation
Correct Answer
A:Private ECR image pulls use supported ECR API/registry interface endpoints and an S3 gateway path for layers; proper DNS/security groups and cost analysis matter.
Incorrect Answers
B:S3 endpoints handle storage-layer service routes but do not terminate ECR authentication and registry API traffic.
C:The ecr.api endpoint alone does not cover the separate ecr.dkr registry endpoint or S3-backed image layer data.
D:AZ-local NAT can address availability and cross-AZ cost but does not fulfill the policy forbidding internet paths or remove NAT processing fees for registry downloads.
Question 369
A software platform expects to retain a stable monthly database spend for three years, but plans to migrate an eligible RDS relational fleet to Aurora and later a portion to DynamoDB. Finance wants a discounted dollar-per-hour commitment that can follow eligible database services and Regions instead of being tied to one RDS DB instance configuration. Which AWS Savings Plans type should it evaluate for supported usage?
- Commit to an EC2 Instance Savings Plan for the current RDS instance family and assume all RDS engine classes are eligible.
- Buy RDS Reserved DB Instances for the existing engine and expect the discount to transfer automatically after the service becomes DynamoDB.
- Database Savings Plans sized to supported eligible database spending, with service and generation coverage verified before purchasing.
- Buy a Compute Savings Plan for the eligible EC2 instances and assume its hourly commitment also covers all Aurora and DynamoDB usage.
Explanation
Correct Answer
C:Database Savings Plans support eligible provisioned and serverless usage across several AWS database services and Regions, subject to current plan eligibility.
Incorrect Answers
A:An EC2 Instance Savings Plan applies to eligible EC2 compute, not to managed RDS and DynamoDB usage as the workload migrates between database services.
B:RDS Reserved DB Instances are a narrower pricing instrument for compatible RDS compute and are not automatically transferable into DynamoDB service charges.
D:Compute Savings Plans cover eligible compute such as EC2, Fargate and Lambda; they are not the database-specific flexible commitment described.
Question 370
A development team has 60 monthly Amazon EBS snapshots from one retired volume. Its finance model predicts that deleting the oldest 50 snapshots will save the entire sum of their apparent logical volume sizes. Compliance permits deletion but requires the newest 10 recovery points to remain restorable. Which cost calculation and cleanup method is technically defensible?
- Multiply the full provisioned volume size by fifty and promise that deleting those snapshots will remove those exact storage bytes regardless of later references.
- Delete the newest ten snapshots first and rely on older snapshots to recover each most recent write generation.
- Copy all retained snapshots to another Region while leaving originals intact, then count the copy as immediate elimination of source-region snapshot charges.
- Identify obsolete snapshots and their exclusively referenced changed blocks, delete only approved recovery points, and measure actual billed storage after retained snapshots continue referencing common blocks.
Explanation
Correct Answer
D:EBS snapshot deletion removes blocks exclusively referenced by deleted snapshots, while shared blocks needed by retained snapshots remain and may be billed against those later snapshots.
Incorrect Answers
A:EBS snapshots are incremental and share underlying block data; deleting an earlier snapshot does not remove blocks that later retained snapshots still reference.
B:Removing the newest required recovery points violates the retention objective; incremental snapshots remain independently restorable while retained, not interchangeable by age.
C:Cross-Region copies create additional stored recovery data; they do not erase original snapshot blocks until actual eligible deletion and reference accounting occur.
AWS Certified Solutions Architect - Professional SAP-C02 Practice Tests