[ale] Microsoft want to know it...

[Chris Ricker]

On Sat, 15 May 1999, Andreas Köpp  wrote:

> http://www.microsoft.com/ntserver/nts/news/msnw/nt4vLinux.asp
> 
> I don´t know what to say about this site...
> ... do they really think they´ve got a chance against a good  installed
> and tuned Linux-System???

Um, look at their test configuration.  I'd be surprised if NT *doesn't* win,
even in the re-match, should it occur.

The major problem with that survey is that it's totally unreflective of
reality.

Nobody runs apache because it's fast.  It's *not* fast.  They run it because
it's portable and is open source (which translates into lots of cool, useful
add-on modules).  At least, that's why I run it on both Linux and Solaris,
even though much faster options are available, including the "free" Sun web
server that ships with Solaris 7.  The logical speed benchmarking, since
that's what Mindcraft is testing, would have been apache on both, or IIS on
NT versus, say, Zeus on Linux.

Similarly, I don't use Linux as an SMB server instead of NT because I think
Linux is faster (though in my case, it may well be, since all the clients
are either Unix or NT).  I use Samba on Linux because I can throw it on an
old pentium in the corner, because I prefer admin'ing Unix, and because
Linux does NFS and AppleTalk out of the box, so I can use it to map all
shares from all systems to all systems (great when you have a mixed network
of NT, various Unix, and MacOS).

Nobody uses that kind of hardware to serve up static web pages, which is all
their web benchmark looks at.  It's orders of magnitude overkill, even for
the relatively bloated NT server they're using.  If you need that sort of
hardware, your web pages are certain to be dynamic and you're dealing with
entirely different bottlenecks than their benchmark measures.

etc.

later,
chris

--
Chris Ricker                                               kaboom at gatech.edu
                                                  chris.ricker at m.cc.utah.edu