AWS Kinesis Outage Affecting Many Organizations

Study how growing a front-end fleet increased resource use on every existing member. The November 25, 2020 Kinesis incident also shows how an intended best-effort telemetry dependency can block a critical service.

Amazon Kinesis supports streaming-data workloads. In the architecture described by AWS’s November 25, 2020 report, a front-end fleet authenticated, throttled, and routed requests to backend shard clusters. The incident affected the Northern Virginia region, US-EAST-1, and services depending on it.

Sequence of events