One might also question the code in the application. Scaling is rarely a simple issue that merely adding a new piece of gear or building a better network can solve. If an application is slow in all of its parts, it's pretty much a hopeless endeavor to just make hardware faster. I don't know anything about Twitters architecture (other than that it is built with RoR, which no one is going to accuse of being fast or proven with regard to scalability). I think it's probably premature to say "the guy failed"...maybe the application or the framework failed. We'll see in 3 months if things have gotten significantly better, or the problems persist.
7 comments
[ 2.0 ms ] story [ 26.3 ms ] threadIt nice to be able to say as it is and not decorated, the guy failed and he left or was made to leave.
lesson learned: scaling is hard!