Designing a distributed rate limiter: algorithms and tradeoffs?

Hey everyone, I'm thinking about how to build a distributed rate limiter for a set of APIs. We need to support various limits (e.g., 100 requests per minute per user, 1000 requests per minute per IP). I've been looking at a few options like the sliding window log and the sliding window counter, and considering how they'd work with Redis or something similar. What algorithms have you found effective for distributed rate limiting, and what were the key challenges or tradeoffs you encountered? I'm particularly interested in consistency vs. performance and how to handle burst traffic effectively.