HP Surestore Disk Array FC60 HP SureStore E Disk Array 12H User's and Service - Page 108

Concepts and Management Optimizing Performance Disk Array Values: Performance Metric 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. Concepts 108