Order events are ingested into Kinesis Data Streams and read by analytics and fraud-detection consumers.
Prevent one consumer's read volume from reducing the other consumer's throughput.
Which two design choices meet the requirement?
Use Enhanced Fan-Out to isolate consumer read throughput and distribute partition keys to avoid hot shards.
Detailed explanation
A balanced key distribution reduces hot shards and uneven ingest or read load.
A balanced key distribution reduces hot shards and uneven ingest or read load.
All records concentrate on the same shard, so scaling benefits are lost.
All records concentrate on the same shard, so scaling benefits are lost.
Each consumer receives dedicated read throughput, reducing contention from shared reads.
Each consumer receives dedicated read throughput, reducing contention from shared reads.
Consumers share read throughput, so contention and latency can increase.
Consumers share read throughput, so contention and latency can increase.
This does not provide independent low-latency stream reads and adds unnecessary reprocessing.
This does not provide independent low-latency stream reads and adds unnecessary reprocessing.
Check official documentation
Compare the following points with the official documentation.
Kinesis Data StreamsのEnhanced Fan-Out、コンシューマー、パーティションキー設計の公式説明を確認する。
What to confirm
コンシューマー分離とシャード負荷分散を別々の制御として選べる。
Key points
- Enhanced Fan-Out
- Partition key
- Hot shard
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
Isolation
Give competing consumers independent capacity when shared polling would create contention.
Distribution
Review key cardinality and traffic skew before deciding how to scale shards.