Strategies for rate limiting across a distributed microservices architecture?

Hey everyone, I'm working on a new system with several independent microservices, and we need to implement effective rate limiting to protect against abuse and ensure fair resource allocation. I'm curious about how others are approaching this in a distributed environment.Are you using an API Gateway for global limits, or pushing rate limiting logic into individual services? What are the pros and cons of each approach you've encountered? Also, for distributed rate limiting, are there any specific algorithms or open-source tools you'd recommend looking into, beyond just token buckets or leaky buckets? Any real-world experiences or pitfalls to avoid would be much appreciated!