Watermark

A stream-processing estimate that event time has advanced far enough to close or update windows while accounting for delayed events.

Why it matters

Watermark helps data teams design systems that are reliable, understandable, governable, and efficient at scale.

Design considerations

• Define ownership, contracts, and expected consumers.
• Make failure, replay, compatibility, and observability behavior explicit.
• Measure quality, freshness, cost, and operational burden rather than optimizing only throughput.