In today's computer systems, the disk I/O subsystem is often identifie
d as the major bottleneck to system performance. One proposed solution
is the so-called redundant array of inexpensive disks (RAID). In this
paper, we examine the performance of two of the most promising RAID a
rchitectures, the mirrored array and the rotated parity array. First,
we propose several scheduling policies for the mirrored array and a ne
w data layout, group-rotate declustering, and compare their performanc
e with each other and in combination with other data layout schemes. W
e observe that a policy that routes reads to the disk with the smalles
t number of requests provides the best performance, especially when th
e load on the I/O system is high. Second, through a combination of sim
ulation and analysis, we compare the performance of this mirrored arra
y architecture to the rotated parity array architecture. This latter s
tudy shows that, 1) given the same storage capacity (approximately dou
ble the number of disks), the mirrored array considerably outperforms
the rotated parity array, and 2) given the same number of disks, the m
irrored array still outperforms the rotated parity array in most cases
, even for applications where I/O requests are for large amounts of da
ta. The only exception occurs when the I/O size is very large; most of
the requests are writes, and most of these writes perform full stripe
write operations.