Ravi, thks for the time in writing this. However I would need some more data for the article to really hit the spot to be honest. For instance, how much does the sql server kernel actually take in memory (just the core functionality disregarding any user db). And how much do instances take? I know it's wrong to assume that 2 instances of the with the exact same data and load will take 2x the memory. And would you be able to quantify how much improvement (if any) can be expected in running sql server as a background process in an OLTP scenario - i really don't see a correlation here you might be assuming something I am not.

Totally agree with you on the collecting data for baselines,

thanks for posting article,

you're a braver man than me,

SNMF