AWS Certified Data Engineer Associate DEA-C01 Practice Question

Kinesis Data Streams in on-demand mode automatically scales shard capacity up and down based on traffic, so the company pays only for the throughput that is actually used during brief spikes instead of permanently provisioning extra shards. A Lambda function invoked by the stream can buffer records and write them to Amazon S3, achieving sub-minute end-to-end latency. Kinesis Data Streams retains data for 24 hours by default, allowing the telemetry to be replayed if needed.

A provisioned stream would need about 25 shards to sustain the 25 000-records/sec peak, so the business would pay for that capacity even when traffic is at the normal 4 000-records/sec level, making it more expensive overall. Running an Amazon MSK cluster introduces higher operational overhead and cost for this throughput, and AWS DMS is designed for database change capture, not high-frequency mobile telemetry ingestion.