When I put more RAM on the box that was running my stress test (as
opposed to the server that is hosting the app) I got approximately a
100% improvement in the speed of my web application!!
This must be incorrect. An improvement to the load simulating machine
shouldn't improve the performance of the application and it leads me to
think I did something wrong. But what?
Note, I watched the load simulating box's processor and it never went
above 50% during the test (either before or after the RAM addition).
It isn't a complicated test. I am only testing a single page and I was
only simulating 10 users.
Before the RAM addition the TTLB was 50 seconds (ugh), after the RAM
the TTLB was 5 seconds!!! I didn't change the server hosting the app
one bit--just the box simulating the load.
The load simulating box (W2k/IIS 5.0) had 128 MB before and 1.28 GB RAM
after. It has a Pentium IV - 1.6 GHZ.
Arrrrghghgh. Thanks!