Great Article. Particularly loved the trace flag tips.

Wondering about two things though:

1. Can we set optimize for ad-hoc workloads simply because we have a large database? I thought this should be based on the "owrkload" actually ... (no pun intended)

2. Because of compatibility issues, I have left some databases in 100 , 110 or 120 (DBMA pointed out potential risks of upgrade). How bad really is this?