This is a project to try and identify suitable tools that will allow us to implement a distributed backup infrastucture. Distributed backup is a step forward from “common” server-based network backup to a serverless distributed backup system that allocates backup data across the whole network or in this case, internet. For the users this would mean donating, say 10Gb to the 'pool' and receiving a similar amount back as distributed storage (it's not quite as straight forwards as that as will be explained at a later date. As it's for personal information, a few criteria should be met.
* All data should be encrypted during transmission
* All data must be encrypted when stored
* An adequate level of redundancy should exist
* A fair method for trading storage and bandwidth must be used
Additionally, any tools must be under a suitable open source licence. It would be great if people could look at projects, evaluate them and document their findings on this page
Distributed Internet Backup System (DIBS) - started in 2002 by Emin Martinian and written in python. It is well documented, installs fine under lucid, works fine on my small intranet of 4 machines, but has no GUI. It also appears to be completely dormant. The code is freely available and the licence appears to be open source compatible.
This is a technically pretty complete, but development seems to have stopped in 2008 - there is a version for gutsy, but that's it!
This isn't really based on P2P backup ideas and is more about producing multiple copies of the same data
Having looked at the site for tahoe-lafs, I'm not sure if it's quite what we want. The others automatically give two-way sharing with the storers; this is only one way. One of the others is probably a better choice.
Distributed storage based on trading local disk space with an option to buy additional space. * http://www.wuala.com/ I've signed up for an account and used it for a while, but find it too slow for my needs and not particularly reliable.
I sent an email to the mailing list in July 2010 for ideas. One reply in particular raised some interesting points https://lists.ubuntu.com/archives/ubuntu-uk/2010-July/025389.html mostly regarding legal issues and moral problems with possibly storing 'inappropriate' material on the network. Current UK legislation requires a computer owner to divulge passwords and encryption codes etc to law enforcement agencies. Quite how this would apply to an encrypted P2P distributed backup solution is difficult to determine. Wuala, the commercial effort listed above, doesn't go into this in any great depth in their documentation.
Many people commented on the desire for an open source version of Dropbox/Ubuntu One, featuring things like version control etc and ?hosted on their own hardware. I've come across couple of projects that are ones to watch - Sparkleshare http://www.sparkleshare.org/ (written in mono) and Persy http://kinkerl.github.com/persy/index.html (written in python). I'm sure there many others out there.