Jonathan Corbet wrote:
>
> I've not done much in the way of rigorous testing, but the LWN code under
> Quixote passes the crucial test: it can handle a full-scale slashdotting
> (on a basic server with a 1GHz processor) without trouble.  That wouldn't
> be possible without a fair amount of optimization work on my part, but it
> shows that Quixote is not getting in the way.
>

I read that in the archives.  I passed that tidbit on to a co-worker.
His comment summed it up, "whoa" .  We have serious servers here
and i'm just not getting the performance from zope that i want.

>
>>Oh what is everyone
>>using as the adapters: SCGI, FastCGI, mod_python, other?
>
>
> mod_python.  I didn't like the FastCGI single-process model when I looked
> at it.  SCGI wasn't really available at the time, so I've not looked at it
> that hard.  The mod_python approach has worked well for me; the only thing
> I've really encountered is that the server should have a lot of memory,
> since the full application gets duplicated in each of (potentially very
> many) apache processes.
>

So with SCGI there isn't the duplication of the application(s) for each
apache process?  That's good to keep in mind.  I'm planning on testing
both out to see how the performance and resource usage stacks up.

Thanks for your comments Jonathan, i appreciate it.

--

Tom Jenkins
Development InfoStructure
http://www.devis.com