Jim,

> >I agree with you that the number of fetches are a etereal measure for

we

> >(me) simple mortals. But as mentioned to track the execution time

will

> >be hard. I don't know about how difficult it is.
> >
> CPU time is impossible to get per thread. Elapsed time isn't good
> enough to be useful.

Don't you think that this is how most users/developers measure things?

My users complain that they are sitting waiting. They don't care if
they have been waiting 15 minutes but only 1 min of CPU time has been
used (because there is a huge number of other query/statements/tasks
running on the server) -- they don't want to be waiting more than xxx
amount of time!

CPU time is only significant for developers only when comparing/ranking
the performance of a set of functions -- CPU time does not have a direct
translation into anything that is consistently meaningful for multi-user
remote server based applications.

This means that the real need is for elapsed time, not number of
reads/fetches, not CPU time.


Sean