how do you handle cache invalidation at scale?

we're having some real headaches with old data messing things up in our cache, especially when the traffic's crazy. we use Redis as our main app cache, and yeah, it's speedy, but keeping it updated with the database is just a pain. anyone else run into this? have you figured out good ways to handle cache invalidation when everything's spread out? any particular tools or tricks you'd suggest, besides just setting a time limit (TTL)