RE: Performance improvement on IBM HTTPServer

If you want to compare performance CGI to FastCGI it's useful to test both
implementations externally (that is how your client's browser sees them). I
have done some simple benchmarking using LWP and two versions of the same
script with different extensions (handlers in Apache). 

# from apache conf file
AddHandler fastcgi-script .pl
AddHandler cgi-script .cgi


# scripts are just sym linked in the file system
/var/www/app.pl
/www/var/app.cgi -> /var/www/app.pl


Then use an external PERL script to benchmark the two...

#!/usr/bin/perl
use strict;
use Benchmark qw(timethese);
use LWP::Simple;

timethese(
    25,
    {
        'cgi' => sub { get('http://hostname/app.cgi'); },
        'fastcgi' => sub { get('http://hostname/app.pl'); },    }
);

Probably something very similar would work for IBM HTTPServer. Hope this
helps.

--Mike

________________________________________
From: fastcgi-developers-bounces+mike=goacolyte.com-xGejAJT2w6yOE5Ap4OspEtHuzzzSOjJt@public.gmane.org
[mailto:fastcgi-developers-bounces+mike=goacolyte.com-xGejAJT2w6yOE5Ap4OspEv8+0UxHXcjY@public.gmane.orget]
On Behalf Of Yves Legault
Sent: Friday, July 22, 2005 1:00 PM
To: fastcgi-developers-xGejAJT2w6yOE5Ap4OspEtHuzzzSOjJt@public.gmane.org
Subject: [FASTCGI] Performance improvement on IBM HTTPServer

I am running fastcgi on an AIX box with IBM's HTTPServer. Everything looks
like it's working fine but now my boss wants me to produce metrics... He
wants to know how much faster the CGI is running and wants to be able to
monitor the performance.
 
Now I know I can look at the access_log to find out when commands have
completed but there doesn't seem to be any indication of how to find when
the request was first received from the user.
 
Does anyone know how to calculate the total time the request is being
manipulated by the serverand application.
 
-- YJRL


___________________________________
fastcgi-developers mailing list
http://fastcgi.com/fastcgi-developers/