Re: Preparations for ZD's upcoming Apache/Linux benchmark
Nathan Hand (nathanh@chirp.com.au)
Fri, 11 Jun 1999 14:21:52 +1000
On Fri, Jun 11, 1999 at 03:10:29AM +0200, Ricardo Galli Granada wrote:
> The proposals (not only yours) are becoming too complex and not affordable
> for a webmaster, who are the target of the khttp. If you do not think so
> (just for benchmarks and hype) do not make things too complex. Just
> implement a very stupid and _simple_ web server.
>
> In a real world (i.e. a real Linux web server), webmaster are running
> _many_ virtual servers (more than twenty) and just small changes in any
> configuration file (i.e., changes in Apache conf and zone names quotes)
> it's a big pain to them/us and I bet most of them (us) are still using old
> srm and access apache configuration files...
>
> You are not only going to force webmasters to change configuration files
> and directories, but also forcing them to teach web designers what a
> static page/image and what's not. Believe me, it's very hard to this kind
> of people to understand this kind of "technical" concepts. Furthermore,
> any medium size website is full of rubish, they are just adding/deleting
> files when a small change is needed (yes, HTML/HTTP entropy sucks).
>
> Users want to answer just yes and not to installation programs, they do
> not want to change everything to get benefits that are not very clear at
> the moment. Indeed, it is cheaper, less time consuming and more
> reliable just to move the disks to a more powerful machine and not to
> waste time changing configurations and content files for every virtual
> server they run (and firewalls).
>
> Be careful, if you force users (who are running hundred of thousands of
> Linux powered web sites) to change everything to gain some CPU cicles,
> they will just move to another OS. Or the new feature will be never used,
> and so becomes in another creep in the ware.
I've been thinking exactly the same thing about all this khttpd stuff.
For the record, I have 2 dozen virtual servers over 3 machines. Some of
the virtual servers share content via NFS. Most of the pages are either
PHP3 or perl based CGI. Images are usually static, but not always (I've
done some horrible tricks with .gif being an executable type, which has
an embedded PHP3/GD script to do on-the-fly watermarking, etc).
So to me this khttpd stuff is useless. I won't be using it. I'll go buy
a beefier machine if content gets too slow, because if it gets too slow
it's because the dynamic content is chewing up cycles, not because I am
dealing with too many hits (I can already handle 100s of hits/minute on
a pentium with dynamic content, that's more than 150,000 hits/day).
This khttpd stuff isn't real world. If it becomes part of apache, and I
don't have to change any configuration files to use it, then it may get
used. But if I have to start separating dynamic/static content, and use
multiple configuration files, or throw away virtual servers, or use non
standard ports, then khttpd just becomes an academic exercise.
--
Nathan Hand - Chirp Web Design - http://www.chirp.com.au/ - $e^{i\pi}+1 = 0$
Phone: +61 2 6230 1871 Fax: +61 2 6230 4455 E-mail: nathanh@chirp.com.au
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/