RE: [Webware-discuss] Zope vs Webware (Not Really)

[Hancock, D. (DHANCOCK) <DHA...@ar...>]

Two cents worth from an not-very-informed user.

We had problems with Zope vs. long-running processes initially and that
drove us to Webware, and we've been happy with it ever since.  With Zope, if
a web request spawned one long-running process, no problem: other web
requests would continue to be answered. As soon as you spawned a second
long-running request, all other web requests would hang until one of the two
long-running processes ended.  We tried the same with Webware, and had no
problems at all handling many concurrent requests, even when some of them
spawned lengthy processes.

It seems to me that adopting a process-based architecture would reduce
performance considerably, but maybe there's some middle ground using FastCGI
or mod_python.  My main concern about mod_python (and mod_perl, which I have
more experience with) is that they bloat up httpd processes considerably.  I
haven't used either with Apache 2.0 in its threaded mode, so I don't know if
it's an issue there or not.

Cheers!
--
David Hancock | dha...@ar... | 410-266-4384


-----Original Message-----
From: Geoffrey Talvola [mailto:gta...@na...] 
Sent: Thursday, June 12, 2003 9:16 AM
To: web...@li...
Subject: RE: [Webware-discuss] Zope vs Webware (Not Really)


pau...@em... wrote:
> One thing that has reportedly been problematic with Zope and
> Webware is the choice of threads for distributing work and
> the scalability issues that this brings with it. Having said
> that, the threaded paradigm is probably so entrenched in the
> average Webware application (or in power-user applications,
> at least) that I can't see a process-oriented variant of Webware any
> time soon. 

I love the coding convenience that Webware's threaded architecture gives me.
It makes caching things like SQL results in memory trivially easy.  It's
memory-efficient.  And so forth.

But there is a drawback.  If there's a bug in any of your C-coded extension
modules that causes the process to crash, it kills the whole app server
including any other requests that were being processed.  And then you lose
all of the current sessions.

This has bitten some people on this mailing list in the past because of bugs
in database adapters.  And right now I'm trying to track down a problem
where I have a servlet that makes an HTTPS request to another web site
(using Python's built-in SSL support), and it looks like every once in a
while, something happens during that outgoing HTTPS request that crashes the
app server.  It only happens like once a week, so I haven't had any luck
tracking this down.  If Webware used some sort of process pool instead of a
thread pool, and served only one request per process, this wouldn't be as
much of a problem.

- Geoff


-------------------------------------------------------
This SF.NET email is sponsored by: eBay
Great deals on office technology -- on eBay now! Click here:
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5
_______________________________________________
Webware-discuss mailing list
Web...@li...
https://lists.sourceforge.net/lists/listinfo/webware-discuss