KrakenD · Rate Limits

Krakend Rate Limits

KrakenD is self-hosted software, not a hosted API service, so it does not impose vendor-side rate limits on its customers. Operators configure their own rate limiting using the built-in `qos/ratelimit/router` (per-endpoint) and `qos/ratelimit/proxy` (per-backend) middleware, plus the Enterprise `security/policies` layer for advanced quotas. License usage is unconstrained — Enterprise pricing explicitly is not linked to API count or throughput.

4 Limits Throttle: 429
Rate LimitingAPI GatewayAPI Aggregation

Limits

License throughput license
requests_per_second
-1
Enterprise pricing not tied to API count or throughput.
Endpoint rate limit (operator-configured) endpoint
requests_per_second
configured per endpoint via qos/ratelimit/router
Proxy rate limit (operator-configured) backend
requests_per_second
configured per backend via qos/ratelimit/proxy
Client rate limit (operator-configured) client
requests_per_second
configured via client_max_rate (token-bucket per client IP / header)

Policies

Operator-defined throttling
Rate limiting in KrakenD is configured by the operator in krakend.json; it is not enforced by the vendor.
Token-bucket
Built-in rate limiters use a token-bucket algorithm with `max_rate` (steady) and `capacity` (burst).
Standard headers
When throttled, KrakenD returns 429 and may emit X-RateLimit-Limit / X-RateLimit-Remaining / X-RateLimit-Reset depending on plugin configuration.
Stateless scaling
KrakenD is stateless; rate-limit state is per-instance unless backed by a shared store. Horizontal scaling effectively multiplies the configured rate per instance.

Sources