Enterprise example · simulator
Per-tenant rate limiter
Send a burst of requests and watch the limiter decide: within budget → served, over budget → 429. Switch the algorithm to feel the difference between a token bucket and a sliding window.
Algorithm
8 req
10 s
4 rps
Allowed
0
Throttled · 429
0
Tokens left
8
Reject rate
0%
Bucket
How each limiter decides
Token bucket
A bucket of limit tokens refills at limit ÷ window per second. Each request spends a token; an empty bucket means 429. Naturally allows bursts up to the bucket size — the choice the design made.
Sliding window
Counts the requests in the trailing window seconds; once that count hits limit, further requests get 429 until the oldest ages out. Smoother, but no burst headroom.
A worked example built through the spectastic enterprise lifecycle — umbrella spec, SLOs, an enforcement floor, and a grounded decision behind every choice.