Hi
> > For a "normal" application it might look like:
> >
> > < 250 TPM: MOSXS G3 450
> > 250 - 1000 TPM Solaris Sparc 5E
> > 1000-7500 TPM High-end Solaris
> > 7500+ TPM Multiple High-end Solaris
Note that most App Server people that quote you TPM figures in their
literature don't use 'normal' apps. They quote apps that are in the
'HelloWorld' variety. A HelloWorld app on a nice 500MHz Pentium II
would do better than a few *thousand* TPM on WO4. But what does that
prove. I've built a more I/O limited app talking to Access and Oracle
and running on a 300MHz Pentium laptop that sustained 500 tpm on WO4. A
app hitting a much vaster database with vaster result sets might only do
half as well (although it might do just as well if you deploy with a few
instances). The point is your manager and the media want to quote a
single figure as THE performance number (so you see outrageous numbers
pop out if only one number is to be provided) but such numbers are
unrealistic. Each problem domain is drastically different. So when you
go comparing tpm numbers around, make sure you compare apples to apples.
What you really want is a spectrum of tpm figures (from HelloWorld to a
Data mining application).
d
