Self-repairing Homomorphic Codes for Distributed Storage Systems - Computer Science > Distributed, Parallel, and Cluster ComputingReport as inadecuate




Self-repairing Homomorphic Codes for Distributed Storage Systems - Computer Science > Distributed, Parallel, and Cluster Computing - Download this document for free, or read online. Document in PDF available to download.

Abstract: Erasure codes provide a storage efficient alternative to replication basedredundancy in networked storage systems. They however entail highcommunication overhead for maintenance, when some of the encoded fragments arelost and need to be replenished. Such overheads arise from the fundamental needto recreate or keep separately first a copy of the whole object before anyindividual encoded fragment can be generated and replenished. There has beenrecently intense interest to explore alternatives, most prominent ones beingregenerating codes RGC and hierarchical codes HC. We propose as analternative a new family of codes to improve the maintenance process, which wecall self-repairing codes SRC, with the following salient features: aencoded fragments can be repaired directly from other subsets of encodedfragments without having to reconstruct first the original data, ensuring thatb a fragment is repaired from a fixed number of encoded fragments, the numberdepending only on how many encoded blocks are missing and independent of whichspecific blocks are missing. These properties allow for not only lowcommunication overhead to recreate a missing fragment, but also independentreconstruction of different missing fragments in parallel, possibly indifferent parts of the network. We analyze the static resilience of SRCs withrespect to traditional erasure codes, and observe that SRCs incur marginallylarger storage overhead in order to achieve the aforementioned properties. Thesalient SRC properties naturally translate to low communication overheads forreconstruction of lost fragments, and allow reconstruction with lower latencyby facilitating repairs in parallel. These desirable properties makeself-repairing codes a good and practical candidate for networked distributedstorage systems.



Author: Frederique Oggier, Anwitaman Datta

Source: https://arxiv.org/







Related documents