Watermark

Definitions

1. A mechanism in stream processing that tracks progress in event time, declaring that no more events with timestamps before the watermark will arrive. Enables the system to finalize window calculations despite late-arriving data. Data arriving after the watermark is either dropped or handled in a side output.

   In plain English: A timestamp marker in streaming systems that says 'we've seen all the data up to this point' so calculations can be finalized.

   Example: "Set the watermark to allow 5 minutes of lateness — events arriving more than 5 minutes late go to the dead letter topic."

Related Terms

• Webhook vs. Polling
• Event-Driven Architecture
• Circuit Breaker
• Buffer
• Chunk
• Windowing
• Backpressure
• Consumer Group
• Chunk Upload