[ Thanks to Adam
Berlinsky-Schine for this link. ]
“I decided to run some tests to see how PHP performs in
various situations. The main purpose of this is to determine
whether or not hitting a database for each page is a serious
performance hit, or whether other factors come into play, such as
CPU time or bandwidth. My initial thought is that the database
will not be a factor, assuming that large amounts of unnecessary
data are not pulled back, and that queries are performed in a
reasonable manner. This document is written as I am running the
tests, so I don’t forget anything (heh). I will make notes and
provide example code, and possibly have a nice test program built
to run all the tests. But in case that is too much work or not
feasible, the code provided here should be enough to at least
duplicate the tests.”
“For starters, I am running these tests on my own workstation: a
Windows 2000 Professional box running on dual Intel Celeron 466’s
with 256 megs of ram. The web server is IIS 5 and the PHP version
is 4.0.0, running as an ISAPI module. I am running the Win32 of
MySQL, version 3.22.34-shareware-debug. I am using IE 5.5 as my
browser, although on some tests such as the output ones, I will
check the results in Netscape 4.08 to make sure there are no large
discrepancies.”
“The actual timing information provided by these tests should
probably not be relied on, since the same machine under Linux would
no doubt perform quite differently. However, the relative timing
data should be fairly valid. One advantage of a dual CPU machine
for running these tests is that the OS (in theory) should not have
to switch back and forth between the web server and the browser
while outputting large amounts of data. This should make the tests
closer to that of a real server environment.”