Hi! Number of servers aren't the problem. Adding more servers isn't a solution (it actually doesn't help at all, and sometimes hurts) and it basically comes down to a few things.

1. Subeta is old. You know this, I've said it a bunch of times. I'm working really hard on modernizing pieces of it at a time. The most recent push here has been the new wardrobe items page which is using a new API built for the wardrobe (that will be implemented on the wardrobe) that is SIGNIFICANTLY faster, and uses way less resources on every query. It's smart, and built on top of laravel who just knows a lot more about architecture than I do. I know it doesn't feel like it, but rewriting the user shops has helped this A LOT.

2. Less relying on things that we don't need, and trying to limit the number of technologies that we use. Today my big code pushes were to move the friend feed and item price data back to MySQL (our database) from MongoDB, a secondary database that we've been using. Mongo is faster, and better at running number queries on data, but it also adds a whole extra database to our stack that has to be maintained, updated, and kept running. Our problems over the past two weeks? All from Mongo issues, which we haven't been able to entirely pinpoint (even with professional outside help). So, the solution is to move back to what we know, and what we've been running for years, and what the rest of Subeta runs on (and we've got a system for backups, etc). The V3 wardrobe uses some Mongo stuff for the backend, but we're moving that away with the new APIs as well (which are much faster without Mongo)

Basically: It's a process. My goal is to be as transparent about this as possible, while still not saying things are "fixed" until they're fixed. The status site has been a big help there because we can just mark things as in progress or investigating, and keep the chain going down until we've resolved the problem (and then another one comes up).

PS: After all of the server changes yesterday, we turned off 50% of our server capacity and the site is still running much better. MongoDB was filling up thousands of requests to those servers and making it impossible for PHP to run. So, in this instance, more servers really wouldn't have helped.