Raid Reliability Considerations - IBM Midrange System DS4000 Series Hardware Manual

Table A-3 RAID levels comparison
RAID
0
1
10
3
5
6

RAID reliability considerations

At first glance, both RAID 3 and RAID 5 appear to provide good protection against drive
failure. With today's high-reliability drives, it appears unlikely that a second drive in an array
will fail (causing data loss) before an initial failed drive can be replaced. But if you look at
RAID 6 and calculate the possibility of data loss, the chance to loose data is theoretically
much less than on RAID 3 and RAID 5.
However, field experience has shown that when a RAID 3 or RAID 5 array fails, it is not
usually due to two drives in the array experiencing complete failure. Instead, most failures are
caused by one drive going bad, and a single block somewhere else in the array that cannot be
read reliably.
Description Application
Stripes data across IOPS
multiple drives. Mbps
The disk's data is IOPS
mirrored to another
drive.
Data is striped across IOPS
multiple drives and
mirrored to the same
number of disks.
Drives operate Mbps
independently with
data blocks
distributed among all
drives. Parity is
written to a dedicated
drive.
Drives operate IOPS
independently with Mbps
data and parity blocks
distributed across all
drives in the group.
Stripes blocks of data IOPS
and parity across an Mbps
array of drives and
calculates two sets of
parity information for
each block of data.
Appendix A. Overview of IBM System Storage DS5000 RAID types
Advantage Disadvantage
Performance, due to No redundancy. If one
parallel operation of the drive fails, the data is
access. lost.
Performance, as Storage costs are
multiple requests can doubled.
be fulfilled
simultaneously.
Performance, as Storage costs are
multiple requests can doubled.
be fulfilled
simultaneously with
better data protection.
High performance for Degraded
large, sequentially performance with 8-9
accessed files (image, I/O threads, random
video, and graphics). IOPS, and smaller,
more numerous
IOPS.
Good for reads, small Writes are particularly
IOPS, many concurrent demanding.
IOPS, and random I/Os.
Good for multi-user Slower in writing data,
environments, such as complex RAID
database or file system controller
storage, where typical architecture.
I/O size is small, and in
situations where
additional fault
tolerance is required.
This is the most reliable
RAID level on the
DS5000 storage
subsystem
513