HP 6100 Software Manual page 27

Identifying virtual disk activity
If you cannot locate a period that meets the necessary I/O load requirements, it may be possible to
identify specific applications that are imposing heavy I/O loads, and suspend them long enough to
perform the upgrade. This can be done by using HP Command View EVAPerf to view virtual disk
activity.
For example, after running HP Command View EVAPerf, you determine that all the virtual disks on
the storage system are nearly idle with the exception of one virtual disk. To perform an online upgrade,
it may be desirable to halt the application that is imposing the load on the busy virtual disk. This will
be far less disruptive to your environment than performing an offline upgrade.
To view virtual disk activity:
1. Double-click the HP Command View EVAPerf program icon on the desktop. The HP Command
View EVAPerf command window opens.
2. Enter the following command to display virtual disk activity:
evaperf vd –cont
Select an appropriate interval and duration that will enable you to view a period during which
you may want to perform the upgrade. For example, the following command will capture data
at 10 minute intervals for 2 hours on an array named EVA1 and store the output in a file named
virtual_disks:
evaperf vd -cont 600 -dur 7200 -sz EVA1 -fo C:\virtual_disks
3. When the command has completed, open the file in an ASCII editor. The virtual disk information
will appear similar to that in
Activity is reported separately for each controller accessing a virtual disk. The total activity for
each virtual disk is the sum of the reported activity for each controller. A virtual disk may also be
a snapshot, snapclone, or a DR group member. In the output, logical unit number (LUN) is used
interchangeably with virtual disk.
Virtual disks must be presented to a host to be seen by HP Command View EVAPerf. However,
replication volumes on the replication system are visible without being presented.
Because the storage system controllers are active/active, one controller is preferred (the owning
controller) but requests can still be processed by the other controller (the proxy controller). In
active/active controllers, all host requests are logged by the receiving controller only, whether
owning or proxy. Thus, all request rate and data rate activity for a virtual disk is the sum of both
controllers.
The statistics you can use to determine the virtual disk activity are listed in
the columns on the screen, left to right.
1. Check the values for requests: Read Hit Request (Req/s), Read Miss Request (Req/s), and
Write Requests (Req/s).
2. Look for values in the hundreds. When the values are in the thousands or tens of thousands,
they indicate a very active process such as an active database that you should not disrupt.
If you see high values for requests during the period you are checking, you should look for
another time period for doing your upgrade.
HP StorageWorks 4x00/6x00/8x00 Enterprise Virtual Array Updating Product Software Guide (XCS
n –dur n –sz array -fo filename
Figure 6.
Table 3, in order of
6.220) 27