Amazon AWS Certified Solutions Architect - Professional SAP-C02 Bundle

This alternative, Buffer accepted work in Amazon SQS and retry with idempotent consumers, is appropriate when A checkout API must accept orders during temporary downstream payment-processor outages. Durable queues absorb downstream downtime without keeping synchronous HTTP requests open.

For the separate situation An inventory event must fan out independently to analytics, billing and notification consumers, Publish to an SNS topic with separate SQS subscriptions for each consumer could be useful. SNS-to-SQS fanout preserves independent subscriber processing and retries.

Make processing idempotent and size the SQS visibility timeout for realistic execution solves the related scenario Workers receive the same message again after a visibility timeout expires; standard SQS provides at-least-once delivery and can redeliver unfinished messages.

Although Publish events to Amazon EventBridge with targeted rules and event buses can be valid when An application emits loosely coupled domain events with multiple routing conditions and SaaS integration, it targets a different technical condition. EventBridge routes matched events to consumers without producer coupling.

Use EventBridge Scheduler to invoke targets at the desired times would address a different requirement: A workflow requires long-running scheduled periodic invocations rather than permanent polling hosts. Managed schedules trigger jobs without maintaining background instances.

This alternative, Configure EventBridge archive and replay where supported before incidents occur, is appropriate when A team needs to replay historical events into revised consumers after fixing a mapping bug. Archived events can be replayed to reprocess matching historical bus traffic.

Although Store an idempotency token and reject already-processed transaction identifiers can be valid when A Lambda subscriber can be invoked twice for one upstream event and must not charge customers twice, it targets a different technical condition. At-least-once event delivery requires application-level deduplication for effects.

Configure EventBridge archive and replay where supported before incidents occur would address a different requirement: A team needs to replay historical events into revised consumers after fixing a mapping bug. Archived events can be replayed to reprocess matching historical bus traffic.

For the separate situation A business workflow calls three services with retries, branching and human approval before final settlement, Use AWS Step Functions to orchestrate the stateful workflow could be useful. Step Functions provides durable state, retry controls and explicit branch transitions.

Configure EventBridge archive and replay where supported before incidents occur would address a different requirement: A team needs to replay historical events into revised consumers after fixing a mapping bug. Archived events can be replayed to reprocess matching historical bus traffic.

For the separate situation An application emits loosely coupled domain events with multiple routing conditions and SaaS integration, Publish events to Amazon EventBridge with targeted rules and event buses could be useful. EventBridge routes matched events to consumers without producer coupling.

Store an idempotency token and reject already-processed transaction identifiers solves the related scenario A Lambda subscriber can be invoked twice for one upstream event and must not charge customers twice; at-least-once event delivery requires application-level deduplication for effects.

For the separate situation A business workflow calls three services with retries, branching and human approval before final settlement, Use AWS Step Functions to orchestrate the stateful workflow could be useful. Step Functions provides durable state, retry controls and explicit branch transitions.

Publish events to Amazon EventBridge with targeted rules and event buses solves the related scenario An application emits loosely coupled domain events with multiple routing conditions and SaaS integration; eventBridge routes matched events to consumers without producer coupling.

Although Use EventBridge Scheduler to invoke targets at the desired times can be valid when A workflow requires long-running scheduled periodic invocations rather than permanent polling hosts, it targets a different technical condition. Managed schedules trigger jobs without maintaining background instances.

For the separate situation An application emits loosely coupled domain events with multiple routing conditions and SaaS integration, Publish events to Amazon EventBridge with targeted rules and event buses could be useful. EventBridge routes matched events to consumers without producer coupling.

Store an idempotency token and reject already-processed transaction identifiers solves the related scenario A Lambda subscriber can be invoked twice for one upstream event and must not charge customers twice; at-least-once event delivery requires application-level deduplication for effects.

Use AWS Step Functions to orchestrate the stateful workflow would address a different requirement: A business workflow calls three services with retries, branching and human approval before final settlement. Step Functions provides durable state, retry controls and explicit branch transitions.

Request the relevant quota increase and design tested capacity fallbacks would address a different requirement: An Auto Scaling group cannot satisfy demand after reaching its service quota. Capacity limits can prevent scaling policies from obtaining required instances.

This alternative, Implement a bounded stale-cache fallback with observability and consistency limits, is appropriate when A cached product catalog may be slightly stale, but checkout must remain available if the catalog backend fails. Graceful fallback preserves read availability without misusing stale data for transactional writes.

For the separate situation A service experiences a sudden 50-fold traffic burst and must avoid exhausting its downstream database, Enforce rate limits, queue buffering and bounded worker concurrency could be useful. Admission controls protect downstream saturation during abrupt bursts.

Increase task redundancy and apply AZ-aware service placement where supported would address a different requirement: An ECS service runs only one task despite a two-AZ target and fails when its AZ is unavailable. Multiple tasks across fault domains prevent one task or zone from being the whole service.

For the separate situation An application scales only when CPU rises, but API latency is already high from request concurrency, Use a target-tracking metric representing request load or concurrency rather than CPU alone could be useful. Workload-aligned scaling responds to the actual bottleneck.

Request the relevant quota increase and design tested capacity fallbacks solves the related scenario An Auto Scaling group cannot satisfy demand after reaching its service quota; capacity limits can prevent scaling policies from obtaining required instances.

This alternative, Use a target-tracking metric representing request load or concurrency rather than CPU alone, is appropriate when An application scales only when CPU rises, but API latency is already high from request concurrency. Workload-aligned scaling responds to the actual bottleneck.

For the separate situation A cached product catalog may be slightly stale, but checkout must remain available if the catalog backend fails, Implement a bounded stale-cache fallback with observability and consistency limits could be useful. Graceful fallback preserves read availability without misusing stale data for transactional writes.

Increase task redundancy and apply AZ-aware service placement where supported solves the related scenario An ECS service runs only one task despite a two-AZ target and fails when its AZ is unavailable; multiple tasks across fault domains prevent one task or zone from being the whole service.

For the separate situation An Auto Scaling group cannot satisfy demand after reaching its service quota, Request the relevant quota increase and design tested capacity fallbacks could be useful. Capacity limits can prevent scaling policies from obtaining required instances.

Although Increase task redundancy and apply AZ-aware service placement where supported can be valid when An ECS service runs only one task despite a two-AZ target and fails when its AZ is unavailable, it targets a different technical condition. Multiple tasks across fault domains prevent one task or zone from being the whole service.

Enforce rate limits, queue buffering and bounded worker concurrency would address a different requirement: A service experiences a sudden 50-fold traffic burst and must avoid exhausting its downstream database. Admission controls protect downstream saturation during abrupt bursts.

Request the relevant quota increase and design tested capacity fallbacks solves the related scenario An Auto Scaling group cannot satisfy demand after reaching its service quota; capacity limits can prevent scaling policies from obtaining required instances.

Although Use a target-tracking metric representing request load or concurrency rather than CPU alone can be valid when An application scales only when CPU rises, but API latency is already high from request concurrency, it targets a different technical condition. Workload-aligned scaling responds to the actual bottleneck.

Enforce rate limits, queue buffering and bounded worker concurrency would address a different requirement: A service experiences a sudden 50-fold traffic burst and must avoid exhausting its downstream database. Admission controls protect downstream saturation during abrupt bursts.