I came across this algorithm while reading the "Design A Rate Limiter" section of System Design Interview by Alex Xu, but I have seen it used many times before. For example, DynamoDB implements RCU's and WCU's with this algorithm.
| Metric | Value |
|---|---|
| Requests Accepted (Last 100 Requests) | 0 |
| Requests Rejected (Last 100 Requests) | 0 |
| Tokens Overflowed (Last 100 Time Units) | 0 |