With an increase in computer technology usage comes an increase in data and the need for secure data storage. Valuable stored data must be kept safe from disk failures, operator mistakes, theft, and even natural disasters. For many organizations, data backup methods remain centralized, expensive, and vulnerable to data loss.

This paper presents Friendstore [1], an efficient backup method with decentralized trust management on peer-to-peer backup systems, where the data is stored on close-proximity trusted nodes only. This system is inexpensive, since the storage contract is established with trusted friends through the usage of wide area bandwidth and the disk space available on the helper nodes in real time. The authors propose a new method of using an XOR(1,2) algorithm, which “doubles the amount of backup information stored at a node, at the cost of transferring twice the amount of data during” the restoration procedure. The system consists of a collection of nodes running the same software that fills mainly “two roles: backing up a node’s local data and helping others store their backups.” Since this system is an online real-time backup system, based on the available space on the neighboring nodes, it consists of four tasks, according to the paper: the storage of local data on remote helpers (backup), periodically checking remote copies of its backup data to see whether the data is intact, creating new copies if the data is not intact (verify and repair), and retrieving the remote backup data following a disk crash (restore).

Currently, the system is written in Java for the Linux, Windows, and Mac OS X operating systems. In order to prepare files for backup on a remote helper, the owner of the data node splits the files into chunks, compresses the data, encrypts it, and replicates it on different helpers within a local area network. The system was evaluated based on trace-driven simulations using about 21 nodes. The experimental results reveal that the actual utilization of trusted decentralized node space is more than 75 percent. Also, when more bandwidth is available, “coding significantly reduces the amount of disk space each node must contribute” toward backup. Finally, the system can back up about 48GB of data with a bandwidth of about 150kbps for a period of five years, with a less-than-0.15 percent loss of data. It is also noted that, if the bandwidth increases, the storage capacity also increases.

Friendstore seems like a viable solution for online, trusted, reliable, real-time, low-bandwidth, high-capacity data backup.