Vandisk: An Exploration In Peer-To-Peer Collaborative Back-Up Storage

As personal computers become an integral part of our daily lives, huge volumes of data need to be reliably managed and archived. Uncorrelated failures within a set of independent personal computers offer the promise of low-cost, reliable data storage. The vanDisk project attempts to realize this promise.The main assumption of our project is that users are willing to donate raw storage space to their peers to increase the reliability of their own data. In our system, users offer a portion of their disks to be used as backup space for other users in exchange for space to store backup copies of their own data, thus decreasing the possibility of catastrophic data loss.A number of characteristics differentiate vanDisk from existing projects that explore this space. First, unlike existing projects that that increase redundancy at the data-block or file level, vanDisk operates at the disk level. This substantially simplifies data management and reduces management overhead at the cost of marginally higher recovery costs from partial failure. Second, all data-related operations are transparently replicated at the data source. Third, our design includes an orthogonal component to manage space and bandwidth.Our system is integrated with Microsoft Windows and offers users a virtual drive that transparently replicates data across multiple machines. As well, a complete, original copy of the data is always available on the user's own system. We have modified TrueCrypt, an open source virtual disk package that offers data confidentiality through encryption, and we have added a new driver layer that redirects and replicates all IO requests to a set of Network Block Device servers offered by the peers to store replicated data. Additionally, we use simple data encoding to offer user-tunable tradeoffs between space overheads, compute overheads, and data reliability.