Regarding Server Downtime Last Friday

 Geomancer by S2Los

We've been having problems since closed beta. We've gotten tips and assistance from mysql experts and such, which is why we've been able to get this far. However, mysql just doesn't have the scalability a decentralized system like cassandra has. It's quite a big change to do, but it's not impossible, as large companies/websites are able to do it. The maintenance was/is to get the first of the cassandra stuff up, or at least to be in the position to start an incremental transition (I'm not exactly sure, as I don't work on that stuff). It obviously didn't go smoothly and we obviously hit a snag or two.

The situation is very similar to Reddit's former problems, but Reddit's problems had something to do with their configuration, I think, not an intrinsic scalability limit. To the community, things will probably be pretty much the same, though. "Ugh, the servers are down" "We're working on it" repeat until we're done or have enough transitioning done for the database servers not to grind to a halt under the load. After we're done, we should be free of the database problems (or at least just be able to add more nodes). However, the current problem isn't something you can just throw more servers at. It will be resolved, though.
A major problem with mysql is that a master/slave setup doesn't work well with the number of writes we need to do (I'm not sure what our current setup is).

Read the full statement by S2IAmRoot via Reddit