SCF Reference Manual for the Storage Subsystem (G06.28+, H06.05+, J06.03+)

Disk Load Balancing on G06.10 and Earlier RVUs
Understanding SAC Ownership
On G06.10 and earlier RVUs each SAC is owned by a processor. A mirrored volume can
communicate with its processors through up to four SACs. These events can cause SAC ownership
to change:
• A hardware failure along an active data path to that SAC.
• A PRIMARY command on a started disk.
• A processor failure or halt on a primary processor.
How a disk process responds to a loss of SAC ownership depends on the RVU:
ResponseRVU
SACS are no longer exclusively owned. This is no longer
an issue.
G06.11 and later
The disk process switches its primary process to the
alternate processor. See “PRIMARY Command Function in
G06.00 Through G06.10” (page 125)
G06.00 through G06.10
The disk process tries to switch to the alternate path. See
“PRIMARY Command Function in G05.00 and Earlier”
(page 128)
G05.00 and earlier
Guidelines for G06.10 and Earlier RVUs
• The disk load balance within an enclosure should be one-sided. Ideally, each enclosure should
have one primary processor that:
◦ Runs all disk processes managing disk volumes in that enclosure
◦ Owns all SACs in that enclosure
• When you use the PRIMARY DISK command and it results in a SAC ownership change, all
the other disks that are actively using that SAC are affected.
Topology Branch Example for G06.10 and Earlier RVUs
The group 03 topology branch includes processors 4 and 5 in the group 03 processor enclosure
and the three I/O enclosures attached to it (groups 31, 32, and 33). The disk processes are
configured as follows:
• Processor 4 owns the shaded SACS and manages the primary disk processes that are running
in the shaded disks.
• Processor 5 owns the white SACs and manages the primary disk processes that are running
in the white disks.
Disk Load Balancing on G06.10 and Earlier RVUs 123