Hacker News Reader

Maybe a difference in performance, but caching comes down to bits-to-be-stored. That likely wouldn't change.

redis does have some built-in features that automatically compact data. [0] I'm sure reddit has considered this, but worth noting.

client-side compression works out better for a couple of reasons:

1. You use less bandwidth in/out of the box. This matters in AWS because your bandwidth is (relatively) limited.

2. Your CPU usage to (de)compress is distributed over a larger fleet. The clients spend the CPU time, not the cache servers.

Perhaps they are already doing it, but couldn't they use client side compression as well?