Date: Tue, 21 Apr 1998 16:49:31 -0600 (MDT) From: Marc Slemko <marcs@znep.com> To: Don Wilde <don@partsnow.com> Cc: advocacy@FreeBSD.ORG, hwg-servers@hwg.org Subject: Re: Freeware Message-ID: <Pine.BSF.3.95.980421164206.16057p-100000@alive.znep.com> In-Reply-To: <353D0853.142D72CA@partsnow.com>
next in thread | previous in thread | raw e-mail | index | archive | help
(infoworld cc removed)
On Tue, 21 Apr 1998, Don Wilde wrote:
> To the Editors of InfoWorld -
> Thank you for the increasing acknowledgement by your magazine of the power and
> value of freeware. My company uses FreeBSD, Apache and Perl, along with other
> freeware, for all its mission-critical internet services, and I'm now expanding
> the freeware presence in our company inside the firewall.
> I would like to propose a challenge. Novell recently crowed about its SPECweb96
> benchmark perfomance. I suspect FreeBSD plus Apache would have little trouble
> trashing either Novell or Microsoft systems, or Solaris on x86 for that matter,
> in the high-performance webserver arena. We are, however, at a disadvantage as
> we don't have the budget to buy ad space and marketing to promote ourselves like
> these huge corporations. Certainly we do more with a lot less people!
I not only suspect that you would be wrong in saying current Apache code
could obtain better results than IIS or Novell's Enterprise port in a
benchmark, I know it. SPECweb results have very limited meaning outside
the "how can we make our software give the highest SPECweb results".
Sure, that is a fun thing to play with and I play with such things with
Apache to the limited degree that my available testbeds and time allows,
but it misses the big picture.
Apache's performance isn't shabby, but it isn't optimized for this sort of
thing. Yet.
> Here's what I'd like to see:
>
> Web Server Face-Off
> ===================
> 1) Identical hardware provided for every participant:
> a) a P100+32MB+2GB-IDE "typical small business" webserver
> b) a P-II+256MB+4x2GB-SCSI "large enterprise" webserver
> 2) A software specification for 2 typical websites built on generic HTML3.2
> capabilities
> 3) 2 weeks to code either CGI or ASP or Java (or any other suitable language) to
> perform the functionality of the spec sites, and set it up on the server as best
> they can in that timeframe
> 4) No modifications allowed to the OS code beyond what is normally allowed to an
> administrator, i.e. 'rebuilding the kernel' is allowed, but not 'tweaking the C
> source' or 'patching the OS', since this is not available to payware users.
> Start with the raw code of the latest release disks available to the public.
> 5) A vendor-neutral test site with testing done on an unloaded 10MBps LAN.
> 6) High-speed Internet access for each on-site team for support and code
> contribution from outside
> 7) A simulated-client-load server (or server set) spitting page requests in a
> controlled manner, the algorithms of which are _not_ disclosed beforehand to the
> participants, nor influenced by any of the participants
> 8) Full disclosure of the tweaks, setups, modifications and code utilized by
> each participant, and the cost of all software utilized in the project
>
> What do you think? I think I've presented a fair assessment mechanism, using
> real-world circumstances. Enough boasting and flaming, let's put it to the test!
I'm not sure what you are trying to assess. This doesn't assess
performance very well, saturating 10 megs is trivial, and it is bogus to
expect a server to perform well against some arbitrary client doing
arbitrary things that don't necessarily have anything to do with the real
world.
Note that I am very aware of the benefits of FreeBSD and Apache over other
solutions, but this really doesn't show them.
--
Marc Slemko | Apache Group member
marcs@znep.com | marc@apache.org
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-advocacy" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95.980421164206.16057p-100000>