On Mon, 2008-10-27 at 13:23 +0100, Jens Vagelpohl wrote:

-----BEGIN PGP SIGNED MESSAGE-----

Hash: SHA1

On Oct 27, 2008, at 13:08 , Roché Compaan wrote:

On Sun, 2008-10-26 at 14:07 -0400, Tres Seaver wrote:

- Plone uses too many indexes, and in particular, uses multiple text

 indexes.  Having extra indexes around "just in case" is a sure lose

 a write time, and may even be expensive at query time (depending on

 the query).

- Particular indexes have performance characteristics based on their

 designed purpose:  for instance, the stock FieldIndex  

implementation

 assumes that the number of documents indexed will be >> the  

number of

 discrete indexable values.  Using such an index in an application

 domain with a very large set of indexable values probably loses,  

and

 in ways which don't show up in early / small-scale testing.

- I'm pretty sure that we haven't yet found the best data structure  

for

 "hierarchy indexes" (e.g., the Plone EPI index, or the stock Zope2

 PathIndex, etc.).  Something like a 'trie' might be optimal for

 pure prefix searching of hierarchies.

- I am confident that the TopicIndex is underutiliized:  it does  

*all*

 the work for a given query at write time, and can thus be  

blindingly

 fast at query time.

- Other special-purpose indexes (e.g., a "recent items" index) would

 be worth a look, especially for applications with large volumes of

 content.

I agree that one should look at improving performance without  

caching as

well. But this is a lot harder and takes significantly more  

development

and debugging time than introducing some form caching. So I'm not

convinced that it needs to happen in a certain order. If caching gives

you lots of performance with little effort now, then why shouldn't you

use it?

It's the typical trade-off. One course is expedient and fast for your  

use case now. The other requires more resources, but benefits  

everyone. Including those who don't want to depend on yet another  

package, like memcached, for performance.

I'm not tied to memcached. We started out using module level caches like
zope.cache.ram but that has obvious problems when using ZEO.

When it comes to integrating anything in Zope itself I'd choose the  

latter.

Sure, we're not trying to get this into Zope, we're just sharing our
experience and exploring the territory so that one can produce a third
party package that really help people with the same use case (which I
suspect is quite common one).

--
Roché Compaan
Upfront Systems                   http://www.upfrontsystems.co.za

_______________________________________________
Zope-Dev maillist  -  Zope-Dev@zope.org
http://mail.zope.org/mailman/listinfo/zope-dev
**  No cross posts or HTML encoding!  **
(Related lists -
http://mail.zope.org/mailman/listinfo/zope-announce
http://mail.zope.org/mailman/listinfo/zope )