how do you deal with cache invalidation when the data keeps changing like its kinda tricky right

hey everyone, been thinking a lot about caching lately, especially when dealing with data that changes frequently. we've got a few services where the underlying data might be updated by different processes, and keeping the cache consistent is proving to be a real pain. we've looked at ttl-based expiration, but that can lead to stale data. event-driven invalidation seems promising, but the complexity of setting up and managing those event streams feels high. anyone have experience with specific strategies or patterns that work well in practice for dynamic, frequently changing data? what are the trade-offs you've encountered