Backpressure and Rate Limiting

Every service has a finite processing rate. If work arrives faster than the service can complete it, the excess has only a few places to go:

It cannot disappear.

A queue can absorb a short burst, but sustained overload makes the queue grow. Latency rises, memory fills, deadlines expire, and retries add even more work. A service that tries to accept everything can eventually complete less useful work than one that rejects excess traffic early.

Backpressure communicates or enforces that an upstream producer must slow down because downstream capacity is constrained.

Rate limiting controls how much traffic a caller or traffic class may send during a period.

They overlap, but they are not identical. Backpressure responds to the consumer's ability to keep up. Rate limiting often enforces a policy or known capacity before overload occurs.