A shard is the unit of capacity in a Kinesis Data Stream. Each shard provides a fixed amount of read and write throughput, so the number of shards determines how much data your stream can handle.
Why shards matter
Each shard supports up to 1 MB/sec (or 1,000 records/sec) write and 2 MB/sec read. If producers exceed a shard’s limit you get ProvisionedThroughputExceeded errors — the fix is more shards (resharding) or switching to on-demand capacity mode, which scales shards automatically. The tell is throughput limits/throttling on a stream.
Test yourself
A Kinesis Data Stream is throwing ProvisionedThroughputExceeded errors during traffic spikes. What’s the best fix without capacity planning?
- Reduce the number of consumers
- Switch the stream to on-demand capacity mode
- Lower the data retention period
- Add a NAT Gateway
👉 Click to reveal the answer & explanation
Correct answer: B. On-demand capacity mode automatically scales shard capacity to match throughput, eliminating manual resharding and throttling. Fewer consumers (A) doesn’t raise write capacity; retention (C) affects replay window, not throughput; networking (D) is unrelated.
Related topics
Kinesis Data Streams · Kinesis Data Firehose · Kinesis Data Analytics
Ready to pass the AWS Data Engineer Associate (DEA-C01)?
Stop guessing whether you’re ready. Our full-length, exam-realistic practice exams put you through the exact question style you’ll face — with a detailed explanation behind every answer, so you learn why, not just what.
- ✓ 6 full-length practice exams
- ✓ A detailed explanation for every single question
- ✓ Realistic, scenario-based questions — not memory dumps
- ✓ Lifetime access, kept current for 2026
Get the DEA-C01 Practice Exams →or try 25 free questions first