## Rebuilding for Array Codes in Distributed Storage Systems

Citations: | 8 - 0 self |

### BibTeX

@MISC{Wang_rebuildingfor,

author = {Zhiying Wang and Alexandros G. Dimakis and Jehoshua Bruck},

title = {Rebuilding for Array Codes in Distributed Storage Systems},

year = {}

}

### OpenURL

### Abstract

Abstract—In distributed storage systems that use coding, the issue of minimizing the communication required to rebuild a storage node after a failure arises. We consider the problem of repairing an erased node in a distributed storage system that uses an EVENODD code. EVENODD codes are maximum distance separable (MDS) array codes that are used to protect against erasures, and only require XOR operations for encoding and decoding. We show that when there are two redundancy nodes, to rebuild one erased systematic node, only 3/4 of the information needs to be transmitted. Interestingly, in many cases, the required disk I/O is also minimized. I.

### Citations

367 |
Polynomial codes over certain finite fields
- Reed, Solomon
- 1960
(Show Context)
Citation Context ...e size. Among coding schemes, maximum distance separable (MDS) codes offer maximal reliability for a given redundancy: any k packets are sufficient to retrieve all the information. Reed-Solomon codes =-=[1]-=- are the most well known MDS codes that are used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD [2] and its extension [3], B-code... |

132 | Row-Diagonal Parity for Double Disk Failure Correction
- Corbett, English, et al.
- 2004
(Show Context)
Citation Context ...wn MDS codes that are used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD [2] and its extension [3], B-code [4], X-code [5], RDP =-=[6]-=-, and STAR code [7]. In an array code, each of the packets consists of a column of elements (one or more binary bits), and the parities are computed by XORing some information bits. These codes have t... |

123 |
EVENODD: an efficient scheme for tolerating double disk failures in RAID architectures
- Blaum, Brady, et al.
- 1995
(Show Context)
Citation Context ...formation. Reed-Solomon codes [1] are the most well known MDS codes that are used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD =-=[2]-=- and its extension [3], B-code [4], X-code [5], RDP [6], and STAR code [7]. In an array code, each of the packets consists of a column of elements (one or more binary bits), and the parities are compu... |

90 | Network coding for distributed storage systems
- Dimakis, Godfrey, et al.
(Show Context)
Citation Context ... can certainly repair this node. But can we transmit less than k packets? More generally, if no more than n − k nodes are erased, what is the repair bandwidth? This repair problem was first raised in =-=[8]-=-, and was further studied in several works (e.g. [9]-[14]). A recent survey of this problem can be found in [15]. In [8], a cut-set lower bound for repair bandwidth is derived and in [11][12][13], thi... |

81 | X-Code: MDS Array Codes with Optimal Encoding
- Xu, Bruck
- 1999
(Show Context)
Citation Context ... well known MDS codes that are used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD [2] and its extension [3], B-code [4], X-code =-=[5]-=-, RDP [6], and STAR code [7]. In an array code, each of the packets consists of a column of elements (one or more binary bits), and the parities are computed by XORing some information bits. These cod... |

52 | MDS array codes with independent parity symbols
- Blaum, Bruck, et al.
- 1996
(Show Context)
Citation Context ...n codes [1] are the most well known MDS codes that are used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD [2] and its extension =-=[3]-=-, B-code [4], X-code [5], RDP [6], and STAR code [7]. In an array code, each of the packets consists of a column of elements (one or more binary bits), and the parities are computed by XORing some inf... |

19 | Deterministic regenerating codes for distributed storage
- Wu, Dimakis, et al.
- 2007
(Show Context)
Citation Context ... less than k packets? More generally, if no more than n − k nodes are erased, what is the repair bandwidth? This repair problem was first raised in [8], and was further studied in several works (e.g. =-=[9]-=--[14]). A recent survey of this problem can be found in [15]. In [8], a cut-set lower bound for repair bandwidth is derived and in [11][12][13], this lower bound is matched for exact repair by code co... |

17 |
An efficient coding scheme for correcting triple storage node failures
- STAR
(Show Context)
Citation Context ...re used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD [2] and its extension [3], B-code [4], X-code [5], RDP [6], and STAR code =-=[7]-=-. In an array code, each of the packets consists of a column of elements (one or more binary bits), and the parities are computed by XORing some information bits. These codes have the advantage of low... |

10 | LowDensity MDS Codes and Factors of Complete Graphs
- Xu, Bohossian, et al.
- 1999
(Show Context)
Citation Context ...are the most well known MDS codes that are used widely in storage and communication applications. Another class of MDS codes are MDS array codes, for example EVENODD [2] and its extension [3], B-code =-=[4]-=-, X-code [5], RDP [6], and STAR code [7]. In an array code, each of the packets consists of a column of elements (one or more binary bits), and the parities are computed by XORing some information bit... |

4 | Existence and construction of capacity-achieving network codes for distributed storage - Wu - 2009 |