Specifications
© IBM Copyright, 2012 Version: January 26, 2012
www.ibm.com/support/techdocs 22
Summary of Best Practices for Storage Area Networks
b) All MDisks within an MDisk Group should be of the same capacity and RAID
level, and in keeping with rule 1, made up of the same kind of disk.
c) Never spread MDisk Groups across multiple back-end disk units, such as two
different IBM 2107 units.
d) If an array has multiple segment sizes to choose from, choose the largest if a
particular MDisk Group will be used for multiple kinds of applications. (This rule
may not apply to some disk arrays, which do not allow the selection of disk
segment size.) The reason for this is that any medium-sized I/O requests
received on the MDisk would then be striped across many platters, robbing them
of needed IOps per Second capacity. The throughput loss through the disk
reading more information than requested is minimal, due to the read speed of
modern disk platters.
e) Deciding how large to make an MDisk Group is a delicate balancing act between
reliability, flexibility, and performance “smoothing”. Larger MDisk Groups are
more flexible and can smooth out I/O load, reducing “hot spots”. However,
striping an MDisk group among a large number of disks makes it statistically
more likely that a given application will be affected in the case of an individual
RAID array failure. At the other end of the spectrum, if each MDisk Group were
made up of just a single back-end RAID array, you lose many of the flexibility
advantages the SVC is designed to provide, and your disk array will likely suffer
from hot spots caused by high-traffic servers.
f) It is important to balance the MDisks among all of the RAID controllers in a
chassis. If a disk array has two controllers (like an IBM DS4000 series), you
would want the MDisks balanced among those controllers. In a DS8x00 series,
you would try to avoid loading up all the mDisks in an MDisk Group on a single
Device Adapter pair.
g) MDisk Groups should contain a number of MDisks equal to a multiple of the
available paths into the disk array. This allows the SVC to provide better load
balancing among the paths.
h) Transaction performance is optimal when each array consists of a single MDisk.
If streamed I/O was dominant, two MDisks per array might be preferred.
Due to the nature of Fibre Channel, it is extremely important to avoid Inter-Switch
Link (ISL) congestion. While Fibre Channel (and the SVC) can, under most
circumstances, handle a host or storage array becoming overloaded, the