Kappa Architecture
Last Updated: May 26, 2026
Kappa Architecture is a stream-first pattern for low-latency processing and historical reprocessing. It keeps source events in an append-only log and rebuilds derived views by replaying that log.
Kappa avoids Lambda's separate batch path, but it depends on durable events, stable schemas, replayable processing, and sinks that can be rebuilt safely.
In this chapter, you will learn:
- What Kappa Architecture is and why it exists
- How replay replaces batch recomputation
- What the event log must guarantee
- When Kappa is better than Lambda