Re: SQL Tuning Regarding System CPU Stats

Thanks Jonathon. I will check out your blog.

> The net result is that multiblock reads are assumed to be more
> expensive than single block reads by a reasonable factor - and
> this tends to be good enough.

This I find interesting. Isn't this the same effect as the OICA parameter? Relative cost of index access vs table scan. What's old is new again.
It sounds like measuring the exact CPU stats is a panacea in theory, but in practice it is not stable enough to rely on, at least not yet. Received on Mon May 21 2007 - 15:59:35 CDT