Real World Webserver Performance

[Ben Laurie <be...@gonzo.ben.algroup.co.uk>]

It occurs to me that the best measure of Webserver performance is to examine
the facts in the real world. The easy way to accomplish this would be to
run a survey. The questions to ask, I imagine, would be:

1. Server s/w, version
2. OS, version, kernel configuration (nonstandard parts thereof)
3. Processor, speed
4. RAM
5. Disk occupied by web pages (size, number of files)
6. Number of hits/day
7. Number of MB/day
8. Processor load (min, max, ave) [if there's a decent way to measure this].
9. Response time for a local minimal page load.

Any thoughts?

Cheers,

Ben.

-- 
Ben Laurie                  Phone: +44 (181) 994 6435
Freelance Consultant        Fax:   +44 (181) 994 6472
and Technical Director      Email: ben@algroup.co.uk
A.L. Digital Ltd,           URL: http://www.algroup.co.uk
London, England.

Re: Real World Webserver Performance

[Brian Behlendorf <br...@organic.com>]

On Wed, 14 Feb 1996, Ben Laurie wrote:
> It occurs to me that the best measure of Webserver performance is to examine
> the facts in the real world. The easy way to accomplish this would be to
> run a survey. The questions to ask, I imagine, would be:
> 
> 1. Server s/w, version
> 2. OS, version, kernel configuration (nonstandard parts thereof)
> 3. Processor, speed
> 4. RAM
> 5. Disk occupied by web pages (size, number of files)
> 6. Number of hits/day
> 7. Number of MB/day
> 8. Processor load (min, max, ave) [if there's a decent way to measure this].
> 9. Response time for a local minimal page load.
> 
> Any thoughts?

Our webserver is current I/O bound, and just about anything you can think 
of affects I/O.  We found out today most of the I/O had nothing to do 
with user-level disk accesses, but instead with file last-accessed time 
updating.  Or so we think.  :)  Every 30 seconds, when the disk cache is 
flushed, we see a huge spike in I/O, which constant running of "sync" 
eliminates, so we'll see tomorrow at 11am PST (our heaviest time and the 
only time we have a problem) if turning that off helps things.  

The point is that there are many other hardware and software related 
variables when looking at server performance.  Any benchmarking suite 
*must* only allow comparisons between the exact same hardware/software 
combination (down to the vendor of the disk controller card) and the 
exact same tests.  Netscape boasts of getting 40 million hits a day, but 
they don't say if that's one box or 40....

	Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com  brian@hyperreal.com  http://www.[hyperreal,organic].com/