Friendster was the extreme case because, of the poor management behind the company. For instance, they went through 4 CEO's over a short period of time and they

Friendster was the extreme case because, of the poor management behind the company. For instance, they went through 4 CEO's over a short period of time and they chose to do a full rewrite from Java to PHP instead of trying to scale the service and because, the rewrite wasn't implemented properly it had to be rewrote again a couple of years later. Moreover instead of sharding/portioning, Friendster chose to scale their service by purchasing an SAN and using daisy-chained MySQL which resulted in considerable slave lag. For comparison, Facebook on the other hand used colleges to "portion" their database in the early stages of the company. Instead of doing the rewrite, Friendster should have focused on trying to scale the service like Twitter did by bringing in PivotalLabs - if Friendster sharded their database for instance it would have helped them immensely and may have resulted in a different outcome for the company.