On Mon, 12 Jul 1999, Indranil Das Gupta wrote:
> PS. Russell, whats your opinion sbout these benchmark reports?
Centered out, eh?
I'm never happy with benchmarks that test very small parts of a system
when overall performance is what interests people. I find it interesting
what the tests concept of a 'small' server is: A Pentium II with 128M of
RAM. I am finally in the last few months deploying servers that large
(EG: There is a K6 and 128M of RAM in the netshooter.com box that hosts
this list) but the vast majority of the LAN servers I maintain are much
smaller than that.
Apache is built for configurability, not for speed. If I were going to
write a web server for Speed I wouldn't do much of what Apache does. Samba
is also not set up to be the fastest it can be, being that it's not Linux
specific and is designed to be portable across POSIX systems. All the
benchmarks tested were these specific applications - one only useful when
talking to Windows boxes (SAMBA) and the other not designed for speed in
the first place.
I'm looking forward to http://www.ottawalinuxsymposium.org/ where one of
the topics will be optimizing server applications for speed under Linux.
Want a real fun benchmark? Try Hoser FTPD vs the IIS-FTPD
http://www.ottawalinuxsymposium.org/hftpd.html
Yet another useless benchmark as I still prefer WU_FTPD due to the level
of configurability of the tool. Benchmarks don't test for
configurability, uptime, servicability and other "full costs of ownership"
issues which are the real core of legitimate platform decisions.
---
Russell McOrmond, Internet Consultant: <http://russell.flora.org/work/>
--
To unsubscribe, send mail to [EMAIL PROTECTED] with the body
"unsubscribe ilug-cal" and an empty subject line.
FAQ: http://www.ilug-cal.org/faq/listfaq.html
