When should you pick a distributed cache instead of a local one for better message queue throughput?
Aidana Tursynbay
·8 views
Hey everyone, I've been thinking about how to speed up our system that handles tons of messages from a queue. Right now, we use a cache right on our worker machines. It helps a lot by cutting down on how often we hit the database. But, as we add more workers, keeping that cache in sync gets tricky. Plus, we start seeing the same data on different machines, which actually makes our cache less useful. I'm wondering if, and when, it makes sense to switch to something like Redis or Memcached. Yeah, you get more network lag and it's a bit more work to manage. But would it be worth it for a better hit rate and more consistent data across all those workers? What stuff do you guys think about when you're picking between a local cache and a distributed one, especially when the main thing you care about is how fast the message queue can go?
0 comments