I'm not convinced we're actually doing this. The only reason I pushed for it is that it felt weird using the file system for caching in production... but everything is going really fast as-is.

The main advantage to Redis (as I understand it) is that it is external storage, so we can share the cache amongst our instances. However the only thing the API server and app server are both doing is very basic things like looking up that project. That being said, Redis might actually count against us because the round trip to the Redis server might be longer than checking the local filesystem.

I tried getting the app/API instances to use Redis a few weeks ago and it didn't work. I didn't really investigate, and personally decided this perhaps wasn't worth the effort.

@Samwilson @Matthewrbowker Thoughts?

I agree that the filesystem cache is probably fine. Like you say, the biggest reason is the fact that we have multiple servers, but the bulk of the cached data probably isn't used by both of them.

We could probably close this declined.

If some 3rd party user of XTools wants to get it fully functional on Redis (if it's not already; because we did have it working briefly when it was on the shared host didn't we?) then it can be looked at again.