Rate limiter

How this question breaks down

1. 1AlgorithmsCore21 minFive ways to decide whether one more request fits the limit, and what each costs in memory, accuracy and burst behaviour.
2. 2DistributedCore26 minEnforce one limit per client across 150 gateway pods in three regions, without adding a slow hop or a single point of failure.
3. 3Client sideCore22 minTell clients their limit before they hit it, reject cleanly when they do, and make sure their retries don't turn one rejection into a stampede.

Work through the parts in order and you’ve covered the full design. Each part is one topic.