Cooperative Regenerating Codes for Distributed Storage Systems
Information Theory
2016-11-15 v2 Distributed, Parallel, and Cluster Computing
math.IT
Abstract
When there are multiple node failures in a distributed storage system, regenerating the failed storage nodes individually in a one-by-one manner is suboptimal as far as repair-bandwidth minimization is concerned. If data exchange among the newcomers is enabled, we can get a better tradeoff between repair bandwidth and the storage per node. An explicit and optimal construction of cooperative regenerating code is illustrated.
Cite
@article{arxiv.1101.5257,
title = {Cooperative Regenerating Codes for Distributed Storage Systems},
author = {Kenneth W. Shum},
journal= {arXiv preprint arXiv:1101.5257},
year = {2016}
}
Comments
5 pages, 7 figures, to appear in Proc. IEEE ICC, 2011