Skip to content

Flow Control vs. Congestion Control

Fast senders must be throttled to avoid overwhelming receivers (flow control) and intermediate networks (congestion control); these are distinct problems requiring different mechanisms - receiver feedback vs. network inference.

Tension Addressed

Simplicity vs. Performance

Recognizable As

Backpressure, rate limiting, windowing, credit-based systems, producer-consumer balancing, queue depth monitoring, reactive streams

Canonical Sources

Jacobson, "Congestion Avoidance and Control" (1988); Reactive Streams Specification (2014)

Transfers To

Stream processing, message queues, API rate limiting, database connection pools, async programming (backpressure)

Sub-concepts: - Backpressure propagation: Slow consumers must signal upstream to slow production - Additive increase, multiplicative decrease: A stable approach to probing capacity without causing collapse