Question on underlying storage of metadata

[orichaud]

I am trying to understand on Alfresco can behave in certain circumstances.
For this, I have created a dummy data model with a single type containing 1,000 properties. And a webscript is in charge, with the Java API on top of the node repository, to fill instances of this type.
After creating thousands of instances of this custom type, the alf_node_properties has grown below several millions rows.
- Will it have an impact on performances?
- Does the cache will still work with such big objects in memory? Does it still really work?

After deleting nodes (call to the deleteNode method), I noticied no change in the alf_node_properties table? Can I force the real deletion of rows connected to deleted nodes?

The rationale for doing this is to be able to store document made of several hundreds of properties distributed in sub nodes connected by child association. Are we not misusing Alfresco?
With such an approach, what will be the behaviour of the form engine? Will it be able to still work ro do we have to look for something more appropriate?

Thanks a lot for your reply and any advice.

Olivier

[mrogers]

Yes the alf_node_properties table does get rather large.   However alfresco anticipates that so it should not be a problem.    
There are instances of alfresco with an order of magnitude more data than yours working just fine.

Cache tuning is always a bit of an art, but yes it does work and makes a fairly significant difference.

After deleting your nodes they were probably moved to the archive store (its a soft delete) As and when you get rid of the nodes then the properties will be removed from the alf_node_properties.

[orichaud]

Thanks for you answers.

How many rows make the difference? I mean, 10 millions or more?

You did not answer to one of my questions: having so many properties for a node (more than 1,000), are we going to kill alfresco?

Regards.
Olivier