We optimized a LOT!

Beforehand and after.

We did patch the code, but we had to guess where the issue was because Google wouldn’t tell us. To this day I don’t know which one of the many fixes we made solved the billing issue… We are not “switching” everytime there is a problem, hell we still have a lot of mission critical App Engine code that is still costing us more than the IaaS equivalent. We’re paying for it every month because we just don’t have the staff to rewrite some of the pieces. We’re doing this migration slowly.

From my discussions with Google it’s pretty clear from the subtext that we would need to migrate anyway as app engine is on the way out (or on the way to a major architectural change). Notice the previous complaints in the thread that our code is ancient 5 year old code… Might as well move to a platform where we have control and visibility. Initially I expected this to be hell as app engine gives so much but experience has shown the opposite. It was MUCH easier working without app engine.

If I understand what you are saying is that you detected all the bugs in your code and optimized every hidden pitfall that might happen in production when you reach millions of users?