IBM ServeRAID, User Reference

Page: 124 / 226

ServeRAID considerations
It is important to understand the following ServeRAID considerations before you set
up your high-availability cluster solution.
Hardware considerations
v You must cable the SCSI channels in a controller pair so that the controllers
share the same channels.
v
You must connect all disks that will be shared to the shared channels. This
includes hot-spare drives, as well as data drives.
v Each shared physical drive must be part of a logical drive or defined as a hot
spare to one or the other servers. You must remove all other physical drives from
the shared disk enclosure to ensure proper operation.
v Do not connect nondisk devices, such as CD-ROM or tape drives, to shared
channels. These devices will be accessible to both servers, and conflicts might
arise.
Controller considerations
v
Controller pair is a term that refers to two controllers, one in each system, that
are connected to each other through the same SCSI bus.
v Each ServeRAID controller must have a different SCSI initiator ID assigned to
the shared channels. For example, set one server, Server A, to ID 6, and set the
other server, Server B, to ID 7. The setup instructions describe this in more
detail.
v You must set the stripe-unit size of each ServeRAID controller in a pair to the
same value (8 KB, 16 KB, 32 KB, or 64 KB).
v
Each ServeRAID controller must have a unique controller name assigned.
v Do not change the controller name on both ServeRAID controllers in a pair at the
same time. Doing so can cause problems and might require that you restart the
server.
v
You cannot share hot-spare drives between controllers in a cluster pairing. If you
want hot-spare protection, each ServeRAID controller must have a hot-spare
drive defined. You must connect hot-spare drives only to shared channels. Both
servers must be able to access their uniquely-defined hot-spare drives in case a
failover occurs
after a hot-spare drive replaces a failed shared drive.
v Physical drives that are configured by one controller in the pair must appear as
ready drives to the other controller in the pair. Be sure to verify this criterion
when you add drives. Restart both servers, or scan for new devices using the
ServeRAID Manager program on both servers, to make sure the added physical
drives appear as ready drives on both servers before creating logical drives.
Array or logical drive considerations
v Create only one logical drive for each array.
v
Every logical drive that is shared by the two servers must have its write-cache
mode set to write through; this will help ensure that data integrity is maintained.
For logical drives that are not shared between the two servers, set the
write-cache mode to write back; this will improve performance.
v You must assign merge-group numbers in the range 1–8 to each logical drive
that will be shared. Merge-group numbers must be unique for each shared logical
drive in the cluster. You must assign merge-group numbers 206 or 207 to the
non-shared logical drives.
108
IBM ServeRAID: User’s Reference, Version 5.10