guideline for name-based web hosting justification

[Kim Scarborough]

> Cool!  Now we all know how to do name based hosting... er, wait... what
> about all those HTTP/1.0 browsers!?  You don't think they exist any
> more?  Check this out.  In fairness I sampled all my virtual hosts off of
> one server from a selective time period.  All my logs files are in the
> www.domain.com format.  Here are my commands and results:
> 
> webserver3: {17} % grep 'HTTP/1.1' www.*.com | wc -l
>   400441
> webserver3: {18} % grep 'HTTP/1.0' www.*.com | wc -l
>   375412
> 
> 48.4% of the browsers out there that accessed my customers' sites used
> HTTP/1.0.  For the uninitiated the 1.0 version of the HTTP protocol does
> NOT support name based hosting.
> 
> Can I tell all my customers to call you when their online business drops
> by almost 50%.  

Wait a minute. When you posted that, I was really surprised. I looked
through my server logs and got similar percentages. Look at this:

/weblogs> grep 'HTTP/1.1' access.log |wc -l
     485
/weblogs> grep 'HTTP/1.0' access.log |wc -l
     449

But guess what. That's a name-based site. It shares an IP with several
sites I host. So obviously, that's not an accurate way to check if people
will load the site--nearly half of this site's hits are from 1.0 requests,
and yet it manages to get the site fine. I'm not sure how this
works--perhaps the browsers are misidentifying the HTTP version in their
requests?--but I can assure you, it does.

I work for an ISP that has hundreds of name-based sites. We haven't got
any complaints from any of our customers about *anybody* not being able to
load their sites in over a year. Saying that 50% of the people out there
can't view name-based sites is just absurd. 

In fact, let's dig a little deeper. Netscape 2.0 and above and IE 3 and
above support name-based hosts. So, since IE 3 reports itself as Mozilla
2, IE 4 reports itself as Mozilla 3, etc., let's try this (on my main,
IP-based site this time):

/weblogs> egrep \(Mozilla/5\|Mozilla/4\|Mozilla/3\|Mozilla/2\) combined.log |wc -l
   74210

/weblogs> egrep -v \(Mozilla/5\|Mozilla/4\|Mozilla/3\|Mozilla/2\) combined.log |wc -l
   6456

So now we're down to 8%. But even that overstates the number of browsers
that can't view name-based sites, because the second number includes
search bots, less-used browsers like Lynx and Opera, and command-line
fetchers like fetch and wget--all of which also support name-based hosts.
I can prune it further upon request. But I would guess it's a fair
assumption that just about the only browsers in use by almost anybody that
can't get to name-based sites are stray copies of Netscape 1.x. So let's
look for that:

/weblogs> grep Mozilla/1 combined.log | wc -l
      79

0.1% of all my hits this month.

So while y'all have a point about the bandwidth accounting, you're on
pretty thin ice when talking about browser incompatability.



--------------73E596B1E8048DB4AD31BD1A--