Monday 11 April 2011 7:02:59 am
Sao, just a few notes if you decide to go with the node implementation... 1) First, in your eZ fetches avoid attribute filtering and sorting at all cost, whenever you can. When using attribute filtering/sorting an sql join is made with two biggest tables, ezcontentobject and ezcontentobject_attribute. And when you don't use attribute filtering, only tables ezcontentobject_tree, ezcontentobject, ezcontentclass and ezcontentobject_name are joined. 2) Depeding on your specific needs you might use eZ Find in order to search through the field that should be searchable. It will be much more faster than doing fetches. 3) In my experience I've done some really nice preformance gains by indexing ezcontentobject by section ID and keeping only a small amount of the newest content in one section (the site has about 10 attributes per object and more than 200 000 objects and growing every day). 4) When importing data, a good thing to do is to skip some parts of the eZ publishing process, see here:
- http://share.ez.no/forums/developer/publish-objects-from-script-without-cache-clear/comment64857
And maybe based on this you could create your own module for publishing your content which also skips some uneeded parts. 5) In eZP 4.5 a new "Asynchronous publishing" feature is introduced, some notes about it here:
- http://share.ez.no/forums/developer/ez-engineering-update-heavy-parallel-publishing-support/
So that might also be worth trying. 6) Database clustering might also be a good idea to try:
- http://doc.ez.no/eZ-Publish/Technical-manual/4.x/Features/Clustering
ezcontentobject_tree, ezcontentobject,ezcontentclass , ezcontentobject_name
--
Nothing is impossible. Not if you can imagine it!
Hubert Farnsworth
|