Order events feed monitoring, aggregation, and auditing applications.
Each application must read the same events independently and replay from its own position after a failure.
Which ingestion platform is most appropriate?
Use Kinesis Data Streams, which retains records and supports independent consumer positions.
Detailed explanation
A message is normally delivered to one worker, so three independent readers require additional fan-out design.
A message is normally delivered to one worker, so three independent readers require additional fan-out design.
CloudFormation manages infrastructure changes and is not an order-event delivery platform.
CloudFormation manages infrastructure changes and is not an order-event delivery platform.
Retained stream records can be read by multiple consumers, each maintaining its own position for replay.
Retained stream records can be read by multiple consumers, each maintaining its own position for replay.
This adds custom batching and coordination and does not naturally provide low-latency independent positions.
This adds custom batching and coordination and does not naturally provide low-latency independent positions.
Check official documentation
Compare the following points with the official documentation.
Kinesis Data Streamsのコンシューマー、保持、再読とSQSの可視性タイムアウトの公式説明を確認する。
What to confirm
ストリームのブロードキャスト的利用と、キューのワーク配布の違いを説明できる。
Key points
- Kinesis Data Streams
- SQS
- Multiple consumers
- Replay
Notes
- Environment: AWS公式DEA-C01試験ガイドと各サービスの公式ドキュメントの確認
- Confirm current AWS service behavior in the official DEA-C01 documentation.
Foundation review
Broadcast versus work queue
Streams let several readers inspect a history; a queue normally assigns each message to one worker.
Checkpoint isolation
Independent checkpoints keep one slow or failed processor from changing another processor's position.