HP Surestore Disk Array 12h - And FC60 Service Manual page 269

Secondary Array Controller
Self Configuring
Self-test
Shutdown
SIMMs
Status Light
Stripe Depth
In disk arrays with two controllers, one controller is called the Primary
Array Controller and the other is called the Secondary Array
Controller. The Secondary Array Controller offers redundant controller
operation should the primary array controller ever fail. The Secondary
Array Controller offers redundant controller operation (full access to all
data) and it also monitors the status of the Primary Array Controller. If
the status of the Primary Array Controller is ever questionable, the
Secondary Array Controller is able to become the Primary Array
Controller. The Secondary Array Controller can also improve the
performance (throughput) of the disk array (host-dependent).
No RAID knowledge is required to use the disk array, since the logistics
of RAID are all managed by the Disk Array Controller.
Self-test is an internal diagnostic test sequence that is performed
whenever the power is switched on. The results of the Self-test, pass or
fail, are displayed by a status light on the front panel of each device.
Shutdown is a coordinated process of taking the disk array offline.
During a Shutdown, vital configuration information is copied from the
array controller NVRAM to the disks. This provides more permanent
storage for this information. In the shut down state, the disk array can
still execute some SCSI commands from the host, but the host cannot
access any data on the array.
The disk array allows a total of three 32-Megabyte SIMMs on the Disk
Array Controller. The maximum capacity is therefore 96 Megabytes.
The SIMMs provide Error Correction Code (ECC).
The Status Light is a light on the front panel of each disk module,
power module, and fan. The status light is used to indicate normal
operating conditions and fault conditions of a particular device.
The Stripe Depth used by the disk array in RAID 5 mode is 64K blocks.
Normally, RAID 5 mode incurs a performance penalty, since in order to
determine the proper parity, previously- written data must be read. This
is called the read/modify/write penalty. However, when the data to be
written exceeds the size of the stripe depth, or 64K, no
read/modify/write penalty is incurred because all of the data (including
parity) already exists in RAM.
Glossary
269