AWS Streams Shard Consumer Lab
Play the same shard lifecycle across Kinesis and DynamoDB Streams to compare discovery, ordering, bounded concurrency, checkpoints, and failure boundaries.
Step0 / 8
Amazon Kinesis Data Streams
Semaphore(maxShardConcurrency) · Sequential polling inside each shard
- 0Ready
- 1Discover shards
- 2Build shard graph
- 3Apply ordering gate
- 4Start shard
- 5Poll records
- 6Deliver downstream
- 7Save checkpoint
- 8Finish shard
Current action
Guaranteed contract
Next transition
Durable KinesisCheckpoint.ShardEnd
Amazon DynamoDB Streams
flatMapMerge(maxShardConcurrency) · Sequential parent-to-child traversal per root tree
- 0Ready
- 1Discover shards
- 2Build shard graph
- 3Apply ordering gate
- 4Start shard
- 5Poll records
- 6Deliver downstream
- 7Save checkpoint
- 8Finish shard
Current action
Guaranteed contract
Next transition
Iterator completion unlocks child traversal
Adapter-owned
│ at-least-once │Bounded pagination and concurrency · per-shard order · cancellation propagation · emit-before-checkpoint · at-least-once resume
Caller-owned
Durable stores · idempotency and deduplication · retry and retention policy · reshard operations · client lifecycle