Plone: High-Performance Indexing using an external non-Zope indexer

Abstract

In standard Plone, when a document is edited, the document gets indexed several times. This is due to the way the indexing in Plone works, and due to some Archetype idiosyncrazies.

This is a performance problem for sites which have either a large number of editing users or large documents to be indexed.

Queued Indexing

By use of collective.indexing, the reindex work is queued (deferred) to the end of the current zope transaction and optimized. Thus no document is indexed twice.

This is a drop-in and forget solution, just add the collective.indexing egg to your buildout.

External Indexing

For faster queries to the index, the Zope index can be replaced with an external indexer which runs in a separate process. There is a ready-made egg and buildout to use Solr as an indexer using several buildout recipes.