TY - JOUR
T1 - Efficient parity placement schemes for tolerating up to two disk failures in disk arrays
AU - Lee, Nam Kyu
AU - Yang, Sung Bong
AU - Lee, Kyoung Woo
PY - 2000/12/31
Y1 - 2000/12/31
N2 - In order to achieve high reliability in disk array systems, two new schemes using dual parity placement, called DH1 (diagonal-horizontal) and DH2 schemes, are presented. Both DH schemes can tolerate up to two disk failures by using two types of parity information placed in the diagonal and the horizontal directions, respectively, in a matrix of disk partitions. DH1 scheme can reduce the occurrences of the bottleneck problem significantly because the parity blocks are evenly distributed throughout the disk array. DH2 scheme uses one more disk than DH1 scheme in order to store the horizontal parities, while the diagonal parities are placed in the same way as in DH1 scheme with a minor change. Even though both DH schemes use almost optimal disk space for storing the redundant information, the encoding algorithms for them are quite simple and efficient. Moreover, both DH schemes can recover rapidly from any two disk failures.
AB - In order to achieve high reliability in disk array systems, two new schemes using dual parity placement, called DH1 (diagonal-horizontal) and DH2 schemes, are presented. Both DH schemes can tolerate up to two disk failures by using two types of parity information placed in the diagonal and the horizontal directions, respectively, in a matrix of disk partitions. DH1 scheme can reduce the occurrences of the bottleneck problem significantly because the parity blocks are evenly distributed throughout the disk array. DH2 scheme uses one more disk than DH1 scheme in order to store the horizontal parities, while the diagonal parities are placed in the same way as in DH1 scheme with a minor change. Even though both DH schemes use almost optimal disk space for storing the redundant information, the encoding algorithms for them are quite simple and efficient. Moreover, both DH schemes can recover rapidly from any two disk failures.
UR - http://www.scopus.com/inward/record.url?scp=0034500115&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0034500115&partnerID=8YFLogxK
U2 - 10.1016/S1383-7621(00)00031-X
DO - 10.1016/S1383-7621(00)00031-X
M3 - Article
AN - SCOPUS:0034500115
SN - 1383-7621
VL - 46
SP - 1383
EP - 1402
JO - Journal of Systems Architecture
JF - Journal of Systems Architecture
IS - 15
ER -