topic Re: Running out of Disk space in Alfresco Archive

Running out of Disk space

aswini — Tue, 17 Mar 2009 09:02:19 GMT

Hi All,I am trying to deploy Alfresco to our company. Recently I discovered that Alfresco takes up enormous space to store the versions of documents. Alfresco seems to store every version of the document as a separate physical file instead of just the changes. This means that soon the disk space wil

Re: Running out of Disk space

mikeh — Tue, 17 Mar 2009 09:40:22 GMT

Any ideas if there's an alternative for this. And if so, how to implement please. - How do I store only the changes of the documents as versions (ideally MS Word/MS Excel)??

Storing deltas for binary file formats is unfeasible really. Also, I'm reasonably sure Microsoft Office doesn't support anything like this natively.

Mike

Re: Running out of Disk space

Tue, 17 Mar 2009 12:56:51 GMT

Huuuum I'm not so sure about that. I worked as a student for 6 months on delta differencing…We found something very interesting and quite advanced, proposed by Google, called "open vcdiff" (written in C++):
http://code.google.com/p/open-vcdiff/

as its name suggests, it's based on the vcdiff RFC (RFC 3284 for those who want to know):
http://www.faqs.org/rfcs/rfc3284.html

We made a little API my partner and I, in C#. It wasn't very hard, although it was only school work (not stronger enough for production and in pre-alpha version should i say ). I think an implementation in Java would be nice

Plus, if I remember, SVN is based on xDelta (not very sure about that) which is also a delta-differencing software?

Maybe an integration of such a thing would be a great idea

Not the good post to speak about that by the way ^^ What do you think of it Mike?

Re: Running out of Disk space

mikeh — Tue, 17 Mar 2009 14:02:35 GMT

Not something on our roadmap as disk space is far cheaper than development effort, but would make a nice contribution!

Thanks,
Mike

Re: Running out of Disk space

Tue, 17 Mar 2009 14:14:13 GMT

I didn't think of it from that angle, but I must admit that I agree
For the contribution, I'll try to motivate my ex-partner and see if we can do something about it…Meanwhile, if anybody is interested, the idea is raised ^^

Re: Running out of Disk space

aswini — Tue, 17 Mar 2009 14:29:48 GMT

So does that mean, document versioning will duplicate a file even if small changes are done (lets say for a 5MB file, making it 10MB for 2 versions)?? And that there isnt any option [yet] to store the changes alone for a document?

Regards
Aswini

Re: Running out of Disk space

zaizi — Tue, 17 Mar 2009 16:13:54 GMT

As the previous post mentions, to store differences only, you need to know specifics for each document format types.

Typically in implementations with versioning, just set up a custom rule to delete,
1. Versions older than specific value
2. All minor versions

This can be configured as per your specific business processes / requirements.