What is a Rate Limit?
A rate limit restricts the requests, operations, or transferred units a client may consume during a defined interval. A service can apply the restriction per user, credential, IP address, resource, or another scope.
How Rate Limits work
A rate limit is the policy or budget, while a rate limiter is the component or service that enforces it by tracking consumption against a key and then admitting, delaying, or rejecting an operation when capacity is exhausted. Fixed windows, sliding windows, token buckets, and leaky buckets differ in how they treat bursts and replenish allowance, so identical headline rates can behave differently at boundaries. In an API or media pipeline, limits protect shared control-plane and processing resources and should be reflected in client scheduling, observability, and retry behavior.
Key facts
- 1A token bucket permits bursts up to its stored-token capacity while enforcing a long-term refill rate; a fixed-window counter can allow boundary bursts across adjacent windows.
- 2HTTP services commonly use status 429 for rate rejection and may include
Retry-After, but providers can define different contracts. Transloadit’s Rate Limiter signals rejection with an error code and a JSON retry delay, so clients should implement that documented behavior. - 3The limit key defines isolation: per-IP enforcement can combine unrelated users behind NAT, while per-credential enforcement lets one leaked or noisy credential exhaust its own quota.
When Rate Limits matter
Design clients to throttle, cache, or batch work before reaching the enforced threshold. On rejection, follow the server’s documented retry guidance and add jitter so concurrent clients do not repeatedly retry together.