How should we lay out the repository?

[darryl_staflund]

Hi everyone,

I am helping a company develop a web application that they will use to upload, categorize, and search corporate documents.  We have developed the content and a a high-level overview of the application design, but are really having difficulty deciding on how to organize these documents in the Alfresco repository.  Millions of documents could eventually be uploaded.

The problem is that there is no thematic 'shape' to the documents that we might use as a basis of organization.  The documents come from different businesses and departments and participate in any number of workflows (outside of Alfresco.)  There is no formal records management in place or any sort of taxonomy that we might use as a basis of organization.  There doesn't seem to be a security model either that we might check in to.  In fact, the existing directories that they use to organize their documents are an amalgam of 3 different filing arrangements, each created by different people for different purposes. Oh, and these documents are close to end-of-life and don't participate in any workflows other than searching and watermarking.

How should we go about organizing these documents?  Is there a tried and tested strategy that has worked well for others?  Should we impose a classification?  If so, based on what?

Thanks everyone.
Darryl

[darryl_staflund]

Hi again,

I have been talking about this problem with a colleague of mine and I think we're going to adopt a strategy similar to Git – where we generate document hash and use it to locate it within the repository.  For example, if the hash value for a given document is:

    12AE45FBBC98847ACCD12E456FB34ECD
‍‍‍

we could store it in:

    /12
        /AE
            /45
                /FBBC98847ACCD12E456FB34ECD (The document name)
‍‍‍‍‍‍

This should ensure even distribution of the document, although we will have to develop a custom search in order to locate these documents.  This is our initial thought, anwyay.

Darryl

[mrogers]

What problem are you trying to solve?   

You could also lay out documents based on creation date,  that's how the content repository works internally.