You can use the IBM® System Storage® Productivity
Center (SSPC)
to monitor key Global Mirror performance
measurements.
It is important to use a Storage Area Network (SAN) performance
monitoring tool to ensure that all SAN components are performing correctly.
This is particularly important when you use an asynchronous copying
solution
such as the
SAN Volume Controller Global Mirror feature.
SSPC monitors
key performance measures and
alerts you when thresholds are exceeded.
Note: If your VDisk or MDisk
configuration
changes, restart the SSPC performance
report to ensure that performance
is monitored for the new configuration.
Use
SSPC to
check the following measurements:
- The Port to Remote Node
Send Response Time measurement is less
than 80 milliseconds. If this measurement is greater than 80 milliseconds
during monitoring, the long-distance link has excessive latency. Ensure
that
the link is operating at its maximum bandwidth.
- The sum of
the Port to Local Node Send Response Time measurement
and the Port to Local Node Send Queue measurement is less than
1 millisecond
for the primary cluster and the CPU Utilization Percentage is below
50%. A
value that exceeds these amounts can indicate that an I/O group is
reaching
the I/O throughput limit, which can limit performance.
- The
sum of the Backend Write Response Time measurement and the Write
Queue Time for Global Mirror MDisks measurement
of the secondary cluster
is less than 100 milliseconds. A longer response time can indicate
that the
storage controller is overloaded.
- The sum of the Backend
Write Response Time measurement and the Write
Queue Time for Global Mirror MDisks measurement
of the primary cluster is
less than 100 milliseconds. If the response time is greater than 100
milliseconds,
application hosts might see extended response times when the SAN Volume Controller cluster
cache is full.
- The Write Data Rate for Global Mirror MDisk
groups measurement of the secondary cluster
indicates the amount of data that is being written by Global Mirror operations.
If this value approaches either the
intercluster link bandwidth or the storage controller throughput limit,
further
increases can cause overloading of the system. Monitor for this condition
in a way that is appropriate for your network.