HP SureStore E Disk Array 12H User's and Service Manual (C5445-90901, September 1999)

Concepts and Management
Optimizing Performance
108
Concepts
Disk Array
Performance Metric
Values:
Typical/Limit Definition
Disk Queue 2-4/>=10 (per disk) Indicates the average number of transactions on all
disk drives waiting in the internal disk SCSI queues.
Too many commands in the queues decrease
performance.
Concurrency >= drive count/
< drive count
Indicates the number of internal simultaneous disk
accesses the disk array is making. The disk array
achieves optimum performance when it is accessing
multiple disks at the same time, so some concurrency
is desirable. Too low a value indicates that the host is
not requesting enough concurrent I/Os to keep the
disk array busy.
Cache Queue 0.05/>=0.32 Indicates the number of commands in the cache
queue. An excessive number may reflect address
conflicts in host I/Os, caused by repeated access to the
same block(s) of data.
Host Queue 0/>32 Indicates the number of host commands waiting in the
host queue. A problem here typically reflects another
bottleneck further along in the controller data path.
For example, if the cache queue is full, commands
will stack up in the host queue waiting for the cache
queue to empty. The counter does not show until 32
commands are in the queue.
Throughput Not Used Indicates the average number of bytes per second
being transferred over the SCSI channel. This metric
reflects channel activity, and does not indicate any
problems with the disk array.
I/O Rate Not Used Indicates the average number of I/Os per second being
serviced by the disk array. This
metric reflects channel
activity; it does not indicate problems with the disk
array.