From flud

flud backup is highly experimental software, and is still in its infancy. Nevertheless, it has achieved some modest success for its ambitious goals, and is currently:

• 100% Decentralized
• 100% Free
• 100% Open

Additionally, flud is making steady progress towards a backup system that is:

• 100% Fair (nodes get as many remote resources as they provide)
• 100% Resilient

This release is appropriate for emulating flud networks on computers you control, or for backing up and recovering data from computers controlled by friends (see Private flud Networks).

Fairness and cheat-resistency have been a high priority since flud's inception, and are critical to making it possible to backup data to anonymous, untrusted peers in a reliable, fair, and resilient way. In this release, flud does not enforce symmetrical storage relationships among nodes, but the architecture and protocol fully support the implementation of that feature, which is planned for the next release (at which time we will make a big push to get some nodes running on the public internet at large). In other words, now is a really good time to give the software a spin.

This release (0.1.1) fixes the following issues from 0.1.0:

• fludrules.init, used as a template to select/exclude a default set of files in $HOME on first run of fludclient, was missing from distributed packages.
• changes in wx (from version 2.6 to 2.8) caused the rendering of the DirCtrlCheckbox custom widget to be too tall and too thin, making fludclient difficult to use on systems with 2.8.

In addition, the flud YUM repository now supports Fedora 8 (previously supported only Fedora 6).

Major new features in the 0.1.0 release included the following:

• Reputation Tracking (nodes develop affinity for reliable peers)
• Better packaging (yum and apt repositories now available)
• Replaced ldpc encoder with zfec encoder, distributed as a standalone package
• Bugfixes

Features:

• Convergent storage (also called single-instance-storage or content-addressable-storage) makes efficient use of storage resources
• Complete recovery of all data requires only that a user provide a single, compact credential (public/private keypair), i.e. as long as the user has this info printed out somewhere, or stored away in an email account, etc, they will be able to completely recover their data

Getting it:

• flud backup is available for unix platforms, and has been tested on several linux variants. rpms, src.rpms, debs, and tarball install are available, as well as svn access (Download).

Stuff that's missing (on the roadmap):

• statistical verify operation
• data expiration
• new protocol command to query for all data owned by a node (dht redundancy)
• samsara-like claims
• plugins to FludScheduler for native file change notifications
• nat-traversal / STUN [1] / STUNT [2]
• miscellaneous: (wiki todo list, bug list)
• testing on and appropriate porting to windows/mac
• protocol negotiations for resources