Data Management for a Distributed Hash Table
This paper presents a new design and implementation of the DHash distributed hash table based on erasure encoding. This design is both more robust and more efficient than the previous replication-based implementation.
DHash uses erasure coding to store each block as a set of fragments. Erasure coding increases availability while saving storage and communication costs compared to a replication based design. DHash combines Chord’s synthetic0 coordinates with the set of fragments to implement server selection on block retrieval.
Experiments with a 270-node DHash system running on the PlanetLab and RON testbeds show that the changes to DHash increase the rate at which the system can fetch data by a factor of six, and decrease the latency of a single fetch by more than a factor of two.The maintenance protocols ensure that DHash is robust without penalizing performance.
KeywordsFile System Hash Table Server Selection Distribute Hash Table Backup System
Unable to display preview. Download preview PDF.
- 1.Planetlab, http://www.planet-lab.org
- 2.Project Iris, http://www.project-iris.net
- 3.Adya, A., Bolosky, W.J., Castro, M., Cermak, G., Chaiken, R., Douceur, J.R., Howell, J., Lorch, J.R., Theimer, M., Wattenhofer, R.P.: Far-site: Federated, available, and reliable storage for an incompletely trusted environment. In: 5th Symposium on Operating Systems Design and Implementation (December 2002)Google Scholar
- 4.Andersen, D., Balakrishnan, H., Kaashoek, M.F., Morris, R.: Resilient overlay networks. In: Proceedings of the 18th ACM Symposium on Operating Systems Principles (SOSP 2001), Chateau Lake Louise, Ban, Canada (October 2001)Google Scholar
- 5.Anderson, R.J.: The eternity service. In: Pragocrypt 1996 (1996)Google Scholar
- 6.BitTorrent website, http://bitconjurer.org/BitTorrent/protocol.html