List:General Discussion« Previous MessageNext Message »
From:Sasha Pachev Date:May 27 1999 4:52am
Subject:Re: Web application benchmark results
View as plain text  
Van wrote:
> 
> Sasha Pachev wrote:
> >
> > I finally got around to installing PHP on our web
> > server, and to celebrate the event, I decided to run a
> > simple benchmark to compare CGI's in C, C++, Perl and
> > SSI, PHP, and raw HTML. I must admit that the benchmark
> > was greatly skewed to favor server parsed languages. The
> > measurement in question was the response time for a
> > request that would send the following HTML:
> >
> > <html><body>Hello,world</body></html>
> >
> > So really the test was about how it it takes to get
> > started, since the amount of time it takes to send the
> > result over 10MB Ethernet is less than 1ms. So here are
> > the results (Linux 2.0.29, Pentium II 233, 64 MB RAM, UW
> > SCSI, Apache 1.3.6 with PHP 3.0.7):
> >
> > Method          Response time (ms)
> > static HTML      2
> > SSI              4
> > PHP              5
> >
> > Statically
> >  linked
> > C CGI            6
> >
> > Dynamically
> > linked C CGI     9
> >
> > Raw Perl CGI     16
> >
> > Statically
> > linked C++ CGI   19
> >
> > Perl CGI
> > with
> > require 'cgilib.pl' 37
> >
> > Dynamically
> > linked C++ CGI   39
> >
> > Again this is all about getting started. Just "Hello,
> > world", that's it. Obviously the results will change
> > quite a bit on a more complex program. I will probably
> > do more tests tomorrow and publish the results. If any
> > body has any suggestions/requests, let me know.
> >
> > --
> > Sasha Pachev
> > http://www.sashanet.com/ (home)
> > http://www.direct1.com/ (work)
> Sasha,
> I'm quite encouraged.  Does this mean I don't have to learn PERL?  >:)
> Thanks for the test.  Much useful.
> Van

Remember, this is just "Hello, world", who can get to the starting line
the fastest type of comparison. Many web applications spend majority of
the response time just getting to the start, and for those this test
indicates that PHP is probably the best solution. However, if there is
any kind of race beyond the starting line, things will change quite a
bit.

-- 
Sasha Pachev
http://www.sashanet.com
Thread
Web application benchmark resultsSasha Pachev27 May
  • Re: Web application benchmark resultsVan27 May
  • Re: Web application benchmark resultsDave27 May
    • Re: Web application benchmark resultsSrfRoG27 May
  • Re: Web application benchmark resultsSasha Pachev27 May
    • Re: Web application benchmark resultsRitschie27 May
      • Re: Web application benchmark resultsDavor Cengija27 May
  • Re: Web application benchmark resultsVivek Khera27 May
    • Re: Web application benchmark resultsRitschie28 May
      • Re: Web application benchmark resultsSasha Pachev28 May
  • Re: Web application benchmark resultsSasha Pachev27 May
Re: Web application benchmark resultsmab27 May
  • Re: Web application benchmark resultsSasha Pachev27 May