From: Rev Simon Rumble [mailto:[EMAIL PROTECTED]]
>
> On Tue, Aug 01, 2000 at 12:16:11PM +1000, Ken Yap uttered:
> > http://www.netcraft.com/survey/
> >
> > And IIS dropped 0.5%.
>
> Of course the methodology is slightly flawed for these sorts of
> comparisons. As anyone who's worked with IIS will tell you, it takes
> many more NT servers to handle the same work load as a bunch of Unix
> Apache servers. Hence there are probably stacks of IIS boxen sitting
> behind individual URLs like hotmail.com (I think they managed to wean
> themselves off Solaris eventually and probably have a cubic kilometre
> of NT boxes to handle the load :)
>
> The methodology, of course, however makes sense when reporting "this
> many sites use this software" as opposed to "there are this
> many public
> web servers running this software".
Netcraft is even more flawed than that. Remember it counts distinct URLs,
not machines. On a large site (www.hotmail.com etc) it will undercount the
number of machines, but on virtual hosted sites it will count the machine
once for each different URL it responds to. Apache has the massive
advantage over things like Netscape in that it is the server of choice for
small, low hit count sites which probably never get more than one or two
hits a month and yet count just as much as slashdot.org as far as the survey
is concerned.
There is another survey which counts number of hits on each type of server
which is probably a better indicator of the popularity of each system. Of
course, there are lies, damned lies and statistics.
John Wiltshire
--
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://slug.org.au/lists/listinfo/slug
