Question 340
A CloudFront-backed site serves large text and JSON objects. Trace data shows network transfer time dominates for clients on slower mobile links, and responses are compressible. Which CDN optimization should be considered?
- Enable Origin Shield for every response while keeping the same uncompressed text payload size for each mobile viewer.
- Increase ALB backend keep-alive and session stickiness while keeping JSON payloads unchanged.
- Enable supported CloudFront automatic compression or origin compression for eligible content types and verify cache behavior.
- Select a different CloudFront price class solely to shrink the uncompressed JSON returned to each viewer.
Explanation
Correct Answer
C:Compression reduces bytes transferred for eligible text responses, improving latency on constrained client networks.
Incorrect Answers
A:Origin Shield can reduce origin misses but does not compress the bytes transferred to a constrained viewer.
B:Connection reuse can reduce setup overhead, but the stated latency is dominated by content bytes and compressibility.
D:Edge-region cost selection does not directly enable content encoding or reduce response size.
Question 341
A distribution serves read-only assets from one S3 origin, and outage telemetry shows intermittent errors from that origin. A second replicated bucket is available and the business wants CloudFront to retry eligible GET requests against it. Which feature is suitable?
- Assume POST and PUT operations always fail over identically to GET through a CloudFront origin group.
- Enable a Route 53 health-checked DNS failover record for a separate distribution but leave the current CloudFront distribution configured with only one S3 origin.
- Configure a CloudFront origin group with origin failover criteria for supported methods, verifying replicas and expected status-code handling.
- Add Origin Shield to the existing single-origin distribution but configure no secondary origin or origin group.
Explanation
Correct Answer
C:Origin failover can send eligible read requests to a secondary origin after defined primary errors, provided the data exists there.
Incorrect Answers
A:Origin failover has method restrictions; write requests require an explicit application continuity design.
B:DNS failover is a separate approach and does not make the current distribution transparently retry eligible GETs at its replicated origin.
D:An intermediate cache layer may mask some misses but cannot select an alternate origin after defined primary 5xx failures.
Question 342
A scientific simulation cluster runs EC2 compute within one Availability Zone. Every job creates and reads millions of tiny, recreatable intermediate objects and end-to-end profiling shows the existing S3 Standard request latency is limiting the run. The application can tolerate loss of this temporary data if the AZ fails and needs consistently single-digit millisecond requests close to compute. Which storage design should it benchmark?
- Use an S3 Express One Zone directory bucket placed in the same supported Availability Zone as compute, testing its session/API requirements and single-AZ availability trade-off.
- Continue with S3 Standard and replicate objects across Regions, expecting replication to lower same-AZ request latency on the active primary bucket.
- Use S3 One Zone-Infrequent Access for all high-frequency intermediate objects solely to lower per-request latency relative to S3 Standard.
- Enable S3 Transfer Acceleration although the EC2 workers and destination objects are co-located in one Availability Zone.
Explanation
Correct Answer
A:S3 Express One Zone targets request-intensive, low-latency access with co-located same-AZ compute and directory bucket semantics, at the cost of one-AZ availability.
Incorrect Answers
B:Replication helps availability of additional copies but does not directly achieve the requested single-digit millisecond requests at same-AZ compute.
C:The infrequent-access class is not the high-request-rate single-digit-millisecond S3 Express design and adds relevant access-cost tradeoffs.
D:Transfer Acceleration targets long-haul transfer to S3 endpoints, not local highly concurrent object request latency.
Question 343
A media company uploads multi-gigabyte packages from overseas studios to one S3 bucket in us-east-1. Multipart uploads and retries are already enabled, but high Internet round-trip times still cause poor transfer performance. The company can change the client endpoint but not the bucket Region. What AWS feature should it test?
- Enable S3 Transfer Acceleration for the bucket and measure end-to-end multipart upload performance through the acceleration endpoint against the existing path.
- Create an S3 Multi-Region Access Point without adding any additional regional buckets and assume its DNS alias brings the overseas studios into the nearest AWS edge automatically.
- Create an S3 gateway VPC endpoint in the us-east-1 VPC and ask overseas studios using the public Internet to upload through that VPC endpoint without private network connectivity.
- Replace the multipart uploader with a single unresumable PUT for the multi-gigabyte packages, assuming retried full uploads are always faster.
Explanation
Correct Answer
A:S3 Transfer Acceleration uses edge ingress and an optimized AWS path for supported long-distance uploads to the existing S3 bucket; improvement should be measured rather than assumed.
Incorrect Answers
B:An S3 Multi-Region Access Point routes to associated regional buckets; simply creating an alias for one unchanged bucket does not provide the specified direct long-haul acceleration endpoint.
C:S3 gateway VPC endpoints serve eligible traffic routed from associated VPC subnets, not unrelated overseas public-Internet clients with no private path into that VPC.
D:A single PUT abandons resumable per-part retries and parallelism, may hit the single-object PUT size limit, and does not improve geographic WAN latency.
Question 344
An RDS application has a growing transactional database. CPU, memory and query plans are healthy, but inserts fail as allocated storage approaches its limit. The team wants storage growth to respond automatically within an approved ceiling. Which RDS feature should be configured?
- Add RDS Proxy connection pooling but leave primary database allocated storage at its near-full ceiling.
- Increase the number of RDS read replicas while keeping the writable primary volume near full.
- Schedule periodic manual RDS storage increments without a maximum-allocation autoscaling policy.
- Enable RDS storage autoscaling with a suitable maximum allocated storage threshold and monitoring.
Explanation
Correct Answer
D:Storage autoscaling can grow allocated storage as the database approaches capacity without manual emergency resize.
Incorrect Answers
A:Connection-pooling changes do not expand database storage for additional inserted rows.
B:Reader capacity does not automatically add writable allocated storage to the primary instance.
C:Manual changes may avert a future incident if timely but do not provide the required automatic growth as used storage approaches the limit.
Question 345
A large Amazon Redshift cluster executes an identical, expensive aggregation query every few minutes against data that changes hourly. The dashboard tolerates a few minutes of staleness and wants lower query latency. Which database optimization is appropriate?
- Create a suitable Redshift materialized view and schedule or control refreshes to match the required freshness.
- Enable query result caching but keep changing SQL comments and nondeterministic expressions so the cache cannot reuse identical aggregation results.
- Increase Redshift concurrency scaling only, continuing to recompute the same joins on every dashboard refresh.
- Increase query timeout so the repeated aggregation is allowed to run longer without precomputing its result.
Explanation
Correct Answer
A:Materialized views precompute repeated aggregation work and can serve recurring reads faster if refresh timing is acceptable.
Incorrect Answers
B:Result-cache reuse requires eligible identical query semantics; changes prevent the intended repeat-query optimization.
C:More concurrent compute does not precompute a repeated expensive aggregate, unlike a materialized view with controlled refresh.
D:A higher time limit does not reduce recurring join/aggregation work or speed repeated reads.
Question 346
A reporting cluster on Amazon Redshift has several highly variable groups of queries. Long batch queries frequently queue interactive dashboard requests behind them, producing poor user experience even when overall throughput is adequate. What concurrency control should be reviewed?
- Tune Redshift workload management query routing and priorities, with concurrency scaling where eligible.
- Set every query to the same queue and priority and assume all workloads have identical latency needs.
- Enable Redshift concurrency scaling without separating mixed workload queues or prioritizing interactive requests.
- Create one materialized view for a subset of batch queries and keep batch and dashboard workloads in the same unrestricted WLM queue.
Explanation
Correct Answer
A:Workload management can separate interactive and batch workloads to reduce queue contention and preserve dashboard latency.
Incorrect Answers
B:Undifferentiated queues preserve the contention between slow batch and interactive work.
C:Concurrency scaling may add compute for eligible demand but does not directly implement the stated interactive-versus-batch WLM isolation strategy.
D:A materialized view can help eligible queries but leaves the general contention between mixed workloads unaddressed.
Question 347
An Athena table over S3 has billions of events partitioned by date. Analysts filter by date but queries still enumerate millions of partition entries through the Glue Catalog and spend time planning before reading relevant data. Which partition metadata optimization should be considered?
- Increase the number of registered Glue partition entries with daily crawlers and expect partition enumeration overhead to disappear.
- Add partition pruning predicates but continue enumerating millions of catalog entries with the same high planning overhead.
- Set an S3 object lifecycle transition on old partitions while preserving the same Glue partition metadata enumeration plan.
- Use Athena partition projection when suitable, with correct projected ranges and storage templates instead of maintaining excessive enumerated partitions.
Explanation
Correct Answer
D:Partition projection can avoid expensive partition catalog listing for predictable partition keys and formats.
Incorrect Answers
A:Adding more metastore partition entries without changing planning cannot remove the catalog enumeration bottleneck.
B:Pruning reduces data scanning, but in the scenario the dominant delay is excessive partition metadata enumeration; projection addresses that class of bottleneck.
C:Storage-class transitions do not change how Athena resolves a large set of partition metadata.
Question 348
A data lake uses Athena for ad hoc queries, but unexpected full-table scans are exhausting the interactive team’s shared query capacity and driving high S3 bytes scanned. Which workgroup control should be evaluated?
- Configure Athena workgroup query data scan limits and cost controls, and require selective partitioned or columnar queries.
- Increase query execution timeouts for all analysts but leave unlimited S3 bytes scanned by each SQL statement.
- Enable Athena scheduled query execution for every unbounded SELECT * and assume scheduling becomes a per-query scan-cost limit.
- Use Athena provisioned capacity solely as a substitute for workgroup per-query data-usage limits.
Explanation
Correct Answer
A:Workgroup limits can prevent extreme scans while data layout and predicates reduce bytes processed.
Incorrect Answers
B:Longer timeouts do not impose cost protections or cap unselective full-table scan volume.
C:Scheduling moves when queries run but does not enforce a bytes-scanned threshold on one query.
D:Dedicated compute capacity addresses compute planning but does not by itself enforce scan-bytes controls.
Question 349
An ecommerce DynamoDB Orders table contains 90 million completed orders and 80,000 pending fraud reviews. Investigators must query pending reviews by reviewer team and review age without scanning completed orders. A review-only partition key attribute is absent from completed items. Which indexing design gives efficient selective queries while limiting index storage and write cost?
- Create a sparse global secondary index with reviewer-team as an index key and review timestamp as its sort key, present only on pending-review items.
- Create a GSI in which the index partition-key attribute is populated with the same value on all 90 million orders and then scan that GSI.
- Enable DAX with no GSI and keep scanning the full Orders table for review status every time.
- Build a local secondary index that can use an unrelated reviewer team as its partition key even when it differs from the base table key.
Explanation
Correct Answer
A:DynamoDB indexes items only when a GSI key attribute exists; keeping that key on pending items yields a sparse queryable index, avoiding indexing completed items.
Incorrect Answers
B:An index populated on every order is not sparse and a single common partition key produces inefficient, potentially hot queries.
C:DAX caches eligible key-value lookups but does not turn a full table scan filtered on pending review into a selective index query.
D:A local secondary index shares the base table partition key; it cannot independently change to a reviewer-team partition key.
Question 350
A service uses an RDS Proxy between Lambda and a relational database. Metrics show many client connections still pinning sessions to database connections because the application relies on session-specific SQL state. Which optimization should be investigated?
- Disable RDS Proxy and assume each connection becomes free to the database merely because it is direct.
- Enable performance metrics for RDS Proxy but leave the application session-level SET variables and temporary tables unchanged.
- Reduce session-pinning behavior where safe and tune connection reuse so the proxy can multiplex more requests onto fewer database connections.
- Open an additional unique persistent database connection for every Lambda invocation and retain all session variables.
Explanation
Correct Answer
C:RDS Proxy pooling benefits can be limited by pinned sessions; changing application session-state usage can improve multiplexing.
Incorrect Answers
A:Removing pooling restores connection churn without addressing the application pinning cause.
B:Observability reveals pinning but does not unpin database sessions or make back-end multiplexing more effective.
D:More pinned dedicated sessions increase the database connection pressure that the proxy is intended to reduce.
AWS Certified Solutions Architect - Professional SAP-C02 Practice Tests