Scale down Kinesis Data Streams on-demand capacity with ODA warm throughput

This article announces the ability to scale down Kinesis Data Streams on-demand capacity using warm throughput, optimizing costs after transient traffic bursts.

• Set lower warm throughput values to trigger capacity reduction on on-demand Advantage streams
• Stream adjusts to requested capacity or peak usage from last hour, whichever is higher
• Reduces excess Lambda invocations and DynamoDB overhead from unnecessary shards
• Use AWS CLI to update stream mode with new warm throughput setting
• Monitor shard count via DescribeStreamSummary API and CloudWatch metrics like IncomingBytes
• Best practice: analyze 24 hours of traffic patterns before scaling down
• Feature available at no additional cost for on-demand Advantage mode streams

Warm throughput scale-down provides elastic capacity management for on-demand streams, releasing excess capacity after traffic spikes while preventing under-provisioning through a one-hour peak safeguard.