Title :
An improvement of quasi-cyclic minimum storage regenerating codes for distributed storage
Author :
Chenhui Li ; Songtao Liang
Author_Institution :
Shanghai Key Lab. of Intell. Inf. Process., Fudan Univ., Shanghai, China
Abstract :
In a distributed storage system, where failure is inclined to happen at one of the storage nodes linked via network, redundancy is often introduced to increase reliability. Erasure coding is universally adopted to optimize the additional storage size for redundancy while enhancing the fault tolerance. Regenerating codes, as a class of erasure codes, make significant contribution to bandwidth-saving when recovering a failed node. A new family of regenerating codes based on quasi-cyclic codes is presented by Bernât Gaston et al. Quasi-cyclic Minimum Storage Regenerating (MSR) codes are very interesting because of their simplicity and low computational requirements. In this paper, an improvement of Quasi-cyclic MSR codes is proposed and its correctness is proven. We not only improve the constraints of construction, but also minimize the number of nodes needed to connect in a repairing process.
Keywords :
cyclic codes; distributed processing; fault tolerant computing; redundancy; storage allocation; bandwidth-saving; computational requirements; distributed storage system; erasure coding; failed node recovery; fault tolerance enhancement; quasicyclic MSR codes; quasicyclic minimum storage regenerating codes; redundancy; reliability; repairing process; storage nodes; storage size optimization; Bandwidth; Educational institutions; Encoding; Joining processes; Maintenance engineering; Redundancy; Vectors; Distributed Storage; Maximum Distance Separable (MDS) Codes; Network Coding; Quasi-cyclic MSR Codes; Regenerating Codes;
Conference_Titel :
Advanced Communication Technology (ICACT), 2014 16th International Conference on
Conference_Location :
Pyeongchang
Print_ISBN :
978-89-968650-2-5
DOI :
10.1109/ICACT.2014.6779130