Backpressure: Why Must Downstream Congestion Reach Upstream?

Backpressure does not mean trapping pressure at the rear. It means allowing an overloaded downstream process to communicate its limit upstream, causing the producer to slow, pause or send less. It addresses a relational problem: when production remains faster than consumption, every component may appear busy while the whole system becomes unstable under a growing backlog.

Consider a producer sending 500 messages each second to a consumer that can process only 100. A buffer can absorb a brief mismatch. If the gap persists, however, the buffer merely postpones exhausted memory and rapidly rising delay. Backpressure lets the consumer state how much it can currently accept, so the producer can adjust its pace and the system can continue operating.

Backpressure is related to buffering and rate limiting, but it is not identical to either. A buffer stores work that cannot yet be handled. A rate limit normally sets a sending ceiling in advance. Backpressure is triggered by the downstream component’s actual condition and travels backwards through the dependency. Its importance lies in recognising that capacity is not an isolated property: what upstream can sustainably produce is determined by what the whole chain can still receive.