PHP performance vs SSI

[Anon]

I work with web sites that use Apache SSI (server side include) statements extensively to generate page headers and navigation. Sometimes the includes nest up to five pages deep. (They consist of mainly if statements and other includes).

I'm programming a new site from the ground up over the next week and I expect to be using PHP for database queries and dynamic pages in the future. So I'd like to use PHP from the ground up so I can plug in these features later.

My question is: If I use PHP statements to perform simple SSI tasks that I'm used to using, will the performance of my site benefit or suffer? Should I stick to SSI and simply use PHP virtual() commands to include headers/footers?

Thanks in advance.

[Anon]

Apache includes a benchmarking utility (ab) that might be informative. I ran a test on my several-years-old K6-300 with 10,000 iterations of an HTTP fetch. The file included four references to a small text file. Here are the results:

virtual() 66.756 seconds

readfile() 57.146 seconds

ssi #include 44.377 seconds

[Anon]

Wow SSI include is %150 faster than virtual()!

Thanks for the benchmark. How fast do you think it can process PHP include()s? What version of PHP did you use? I plan on using PHP4 with the Zend optimizer.

Thanks again for the info. You put up numbers against others that tell me that PHP should outperform SSI.

[Anon]

3.0.16 with PHP implemented as a module (Linux Mandrake 7.1, 2.2.15 kernel, Apache-AdvancedExtranetServer/1.3.12). X and the usual background junk running.

This test doesn't necessarily mean anything in a real world context -- unless you're in the habit of creating pages that look like this:

<? include("sometext.text"); ?>
<? include("sometext.text"); ?>
<? include("sometext.text"); ?>
<? include("sometext.text"); ?>