Well, I have had a lot of fun using Twitter over the last few weeks. Not only have I found some fun people to follow (no close friends yet… until Jvanti starts updating!) but updating from anywhere via text message is pretty cool. I can post an update on my way home from work and let people know what game I’m planing on playing that night. It’s all great… when it works.
But lately it seems Twitter has been experiencing some problems. With some serious technical hurdles, they seem to be going down for small amounts of time at least once a day. Lot’s posts have been popping up regarding their frustrations with the service. I was beginning one of those “down with Twitter” posts when I started writing this article… until I found out some interesting things about the service. While some of these facts are history (they are upgrading the service as we speak) they can still affect Twitter and are interesting (on an Uber-Mega-Geeky level!)
- Twitter runs (ran?) off of a single MySQL server replicated onto two slaves. I don’t even really know what that means but it sounds really hard.
- Did you just say 3 machines run Twitter? Yes. Used by lots of people, you’d expect a little more hardware.
- If one of those servers goes down, it is caught by a real person, monitoring the database, and is switched and repaired by hand. Then the failed database is rebuilt. This can shut the site down for up to half an hour.
- Someone (no names were mentioned) was responsible for some major outages. Internal drama but the Twitter team is quick to point out “Twitter is a team—we share responsibility for our victories as well as our mistakes.” It is mentioned this person is a former employee though.
So the next time you see this image with the notice:
be thinking about the poor guy on the other end, trying to rebuild that database.
[Photo by Florian]