Having done some consulting for Yahoo, they have some freaking brilliant people. It also means that they tend to pick the best tool for the job. MySQL doesn't scale to 2 petabytes ... apparently neither does Oracle. Which means what? That a 2 petabyte database handling 1M events/hour requires something extremely specialized. That specialization generally requires a couple developers who know a codebase really really well. That could just as easily have been a new MySQL engine.
MySQL was used as the platform for Kickfire, SQL on a chip. $800M in VC funding from Sun and other players. They have a bad idea?
I'm not saying one's better than the other. They have different benefits and different drawbacks.
My criteria for good software:
Does it work?
Can someone else come in, make a change, and be reasonably certain no bugs were introduced?