Home // International Journal On Advances in Networks and Services, volume 11, numbers 3 and 4, 2018 // View article

Reliability Evaluation of Erasure Coded Systems under Rebuild Bandwidth Constraints

Authors:
Ilias Iliadis

Keywords: Storage; Reliability; Data placement; MTTDL; EAFDL; RAID; MDS codes; Information Dispersal Algorithm; Prioritized rebuild; Repair bandwidth; Network bandwidth constraint.

Abstract:
Modern storage systems employ erasure coding redundancy and recovering schemes to ensure high data reliability at high storage efficiency. The widely used replication scheme belongs to this broad class of erasure coding schemes. The effectiveness of these schemes has been evaluated based on the Mean Time to Data Loss (MTTDL) and the Expected Annual Fraction of Data Loss (EAFDL) metrics. To improve the reliability of data storage systems, certain data placement and rebuild schemes reduce the rebuild times by recovering data in parallel from the storage devices. It is often assumed that there is sufficient network bandwidth to transfer the data required by the rebuild process at full speed. In large-scale data storage systems, however, the network bandwidth is constrained. This article obtains MTTDL and EAFDL of erasure coded systems analytically for arbitrary rebuild time distributions and for the symmetric, clustered, and declustered data placement schemes under network rebuild bandwidth constraints. The resulting reliability degradation is assessed and the results obtained establish that the declustered placement scheme offers superior reliability in terms of both metrics. Efficient codeword configurations that achieve high reliability in the presence of network rebuild bandwidth constraints are identified.

Pages: 113 to 142

Copyright: Copyright (c) to authors, 2018. Used with permission.

Publication date: December 30, 2018

Published in: journal

ISSN: 1942-2644