Lambda Architecture

Lambda Architecture combines two processing paths over the same source data: a batch layer that periodically recomputes authoritative views and a speed layer that processes recent events with low latency. A serving layer merges their outputs for queries.

Why it exists

The pattern was designed to balance low-latency results with the correctness and recoverability of full batch recomputation. Immutable source data makes it possible to rebuild derived views when logic changes or errors are discovered.

Trade-offs

Maintaining equivalent logic in batch and streaming systems is expensive and can produce inconsistent results. Modern stream processors can replay durable logs and support stateful computation, so many systems prefer simpler streaming-first or lakehouse approaches. Lambda Architecture remains useful as a mental model for separating fast provisional results from slower authoritative recomputation.