“The events that hit our system the hardest are generally when â€œpopularâ€ users – that is, users with large numbers of followers and people theyâ€™re following – perform a number of actions in rapid succession.”
The Scobleizer isn’t taking all this well, however. On FriendFeed (which seems to be his new social network of choice), he says that Twitter blaming him is “bulls**t,” and that the service was having problems long before he came along with his thousands of friends:
FriendFeed is 1000 times more reliable. Twitter was going down before I even got popular on the service. Their architecture has always sucked and everyone knows it. They’ve never been able to get a handle on the quality of their service and now it looks like they are blaming their top users.
To be fair, the comments from Alex Payne don’t seem to be pointing the finger of blame. It sounds like a simple explanation of why the system goes down so much. Although he says Ruby isn’t to blame, it seems fairly obvious that the service’s architecture has had “scaling” problems, where handling hundreds of thousands of events ties it in knots. As MG Siegler notes in the VentureBeat post, Twitter has to effectively rebuild its system while it is still running — something that other companies have described as “repairing an airplane in mid-air.”
Maybe the $15-million that the company is said to have raised from Spark Capital and other venture funds will help with that task. So should Twitter limit the number of friends you can have, the way Facebook does? Some people seem to think they should.