Hitachi MK-96RD617-08 - Manual (Page 27)

About Performance Manager Options 2-11
Hitachi Universal Storage Platform V/VM Hitachi Performance Manager User’s Guide
Hard Disk Drive Workload Statistics
If particular hard disk drives or data are heavily accessed, system performance
might deteriorate. Performance Monitor lets you view statistics about parity
groups and logical devices to help you detect bottlenecks in your system. If
you mapped volumes in an external storage system, Performance Monitor can
also monitor the access workloads of the external volume groups and the
external volumes. Performance Monitor displays a line graph indicating
changes in access workloads, so that you can detect the peak I/O access
times.
You will be unable to view workload statistics that expired a specific storing
period because such statistics are erased from the storage system. The storing
period of statistics is only short range (between 8 hours to 15 days) and that
changes depending on the collecting interval and the number of CUs to be
monitored specified by the user. For details on the storing period of statistics,
see Understanding Statistical Storage Ranges.
Workload information mainly displayed by Performance Monitor is as follows:
• I/O rate
The I/O rate indicates how many I/Os are made to the hard disk drive in
one second. If the I/O rate is high, the hosts might consume a lot of time
for accessing disks and the response time might be long.
• Transfer rate
The transfer rate indicates the size of data transferred to the hard disk
drive in one second. If the transfer rate is high, the hosts might consume a
lot of time for accessing disks and the response time might be long.
• The read hit ratio
For a read I/O, when the requested data is already in cache, the operation
is classified as a read hit. For example, if ten read requests have been
made from hosts to devices in a given time period and the read data was
already on the cache memory three times out of ten, the read hit ratio for
that time period is 30 percent. A higher read hit ratio implies higher
processing speed because fewer data transfers are made between devices
and the cache memory.
• The write hit ratio
For a write I/O, when the requested data is already in cache, the operation
is classified as a write hit. For example, if ten write requests were made
from hosts to devices in a given time period and the write data was already
on the cache memory three cases out of ten, the write hit ratio for that
time period is 30 percent. A higher write hit ratio implies higher processing
speed because fewer data transfers are made between devices and the
cache memory.