[Zope] performance

[Paul Zwarts]

Lars,

Cant say about How-Tos on the subject otherwise i wouldnt have asked ; ) ,
but a discussion could be started on how a product could be created that did
this. I've for a while envisioned a product that would create a ZClass
through variables. You are supplied with a management screen to create SQL
tables by using drop down lists and a 'number of fields' field. This product
would then create the SQL database and simple ZSQL methods for retrieving
and adding. (I once made a clunky but powerful generic dtml method that
checks security and thus not needing to custom make ZSQL methods).

>From this discussion I see another option could be that this product creates
a SQL index specificaly for indexing Zope objects that contain
primary/foreign key relations to the database. This allows SQL to be used
for content indexing (CPU intensive, but not client download intensive)
actions, and allow Zope to store the actual HTML. Any further use would have
to be customizable, but at least a datamodel creation tool would quickly
help alot of zopers.

I find that using SQL exclusivley is more work than the performance gain,
and content in SQL is even worse because you cannot index or search or crawl
it.

Paz

-----Original Message-----
From: Lars von Wedel [mailto:vonWedel@lfpt.rwth-aachen.de]
Sent: Thursday, July 12, 2001 12:01 PM
To: Paul Zwarts
Cc: Zope
Subject: Re: [Zope] performance



Hi Zopers,

That's generally an interesting discussions for all Zope applications
which are somewhere between low- and high-volume data amount. I suppose
that simple, low-volume data is best handled by some Python objects
stored
in ZODB.

But where are the limits of internal features such as ZCatalog, for
instance?
Do they keep up with indexing tools such as mg?

> Perhaps a dual-usage schema could be used. Objects are created using
> ZClasses and reside in Zope (for crawler indexing and client bookmarking,
> and no query strings) and each has its index duplicated in Zope for
> application indexing purposes. The actual Zope object wouldnt be requested
> until its page was needed, and not its presence in an index. I think that
> would be the best of both worlds. Comments, or did I just solve my own
> problem?

I like the approach, thought about something like that too -- kind of a
proxy
in Zope and the real data in some external database. I'd appreciate if
someone
could point me (and others) to some docs or howto's explaining more
details
on this problem!

Lars