Document management system: what to use as storage backend (docs content repository)?

Question

I want to make a document management system (interface in Ruby).
What do profesional sollutions (Alfresco, Liferay social office, others) use for storing and versioning documents?
What else can I use?

Key points:

• storage space optimization (deltas, compression ...)
• versioning
• ability to index docs (can be external)
• ability to make backups at runtime (live hot-backup)
• locking?
• scalability on large data volume
• ensure data integrity (hashing?)
• permissions
• transactional
• Workflow support (optional)

Bonus points:

• how does KnowledgeTree do it?
• how does Liferay Social Office do it (jcr?) ?
• how does Alfresco do it ?

Any books on this issue ?

Answer 1

Most of the enterprise document management solutions I've seen (Cimage, Documentum, LiveLink) definitely don't care about #1. Storage is relatively cheap, especially if it's storage vs processing (store and retreieve). They mostly rely on filesystem based storage - perhaps with name abstraction such that ShoppingList.doc perhaps becomes 20100909100101a.doc.rev1, with a database tracking the given-name, the stored name, revisions, and various other data {MIME type, headers & properties etc}. By not generating deltas + compression you get indexing very easily from any number of existing products/agorithms. Versioning is also extremely simple with this approach.

Depending on the size and scale you're building, you could also store versioned files within a database.

An (S)FTP or CIFS storage process would also allow your software to run on an app server with modest space, but store the files+history on a file or cloud server of some sort - although this isn't much different from filesystem based storage.

Answer 2

Y'know, my first instinct would be to just use Subversion. You'll need external indexing, and if you want to store hashes you'll need to write some code to do it yourself, but everything else fits.

It ships with ruby bindings, though I'm not personally familiar with their quality.