Practical limit on how many documents to store in a space?

[stevegreenbaum]

We are implementing a RESTful content management service for a case management system and had a question about storing all of the documents in a single space under Company Home.  I have read online that there is a performance concern about storing too many documents in a single space.  However, I can’t tell if this is a concern with how the Alfresco OOTB user interface performs with many (I think it mentioned 2000) documents in a single space or is this a concern with how the repository will perform? 

Should we store all of the case documents in a single space under Company Home or do we need to create a hierarchy of spaces to limit how many documents are in any one space?  We will never be using the Alfresco UI for this application?  We are writing a custom UI that will retrieve approx. 20 documents at a time that are related to a single case.  Most of the queries will be to retrieve all documents related to a single case or to retrieve a single specific document. This repository will have 50M+ base nodes plus renditions. 

Is there a practical limit on how many documents that should be stored in a single space given our use-case?  I am looking for the technical rationale for why create a space hierarchy might be helpful in our case.  Thanks

[mrogers]

It depends upon how your UI is going to retrieve documents.   If through the file path then there may be issues,  if something else then the path is irrelevant.

Share should now be "paged" anyway so former issues with displaying too many documents in one folder should not matter.

[stevegreenbaum]

Ok, thanks.

We are planning to use CMIS and selecting documents using a property that has been indexed (e.g., caseId).  From your post I think we should be okay, correct? 

When you say "If through the file path then there may be issues", I think you are referring selection using a technique such as childByNamePath (string path) or childrenByXPath (string xpath), correct?  I assume that those retrieval techniques don't use an index.