what's your go-to for rate limiting in distributed systems

Hey everyone, I'm trying to nail down a solid strategy for rate limiting across our microservices. We've run into some problems with thundering herd and making sure everyone gets a fair shake. I've checked out token bucket and leaky bucket algorithms, but getting them to work right and scale up in a distributed system seems pretty complicated. What tools or patterns have you guys found work well for handling API rate limiting when you've got a bunch of services? Any pitfalls I should watch out for?