1.9.5 Monitoring disk resources

This subsection explains how to monitor the disk resources of a logical partitioning feature system.

Organization of this subsection

(1) Overview
(2) Monitoring examples
(3) Customizing monitoring template reports

(1) Overview

Each LPAR on a system with logical partitioning feature uses the SAN boot to start. The system with logical partitioning feature assigns the HBA installed on the physical host of the host machine to each LPAR as a virtual HBA. The LPAR works recognizing the virtual HBA as a physical HBA.

PFM - RM for Virtual Machine obtains the number of I/O interrupts on the physical and virtual HBAs as performance data. If this value remains high, it is considered that the HBA is overloaded by the LPAR, and thus you can take an appropriate corrective action.

Note that PFM - RM for Virtual Machine cannot monitor whether there is a lack of physical disk resources on the SAN storage that exists for each LPAR. Therefore, you must monitor it by using PFM - Agent for Platform or PFM - RM for Platform.

Also, if you want to detect lack of disk resources, failure in HBAs, and other errors, you can use alert monitoring through the LPAR Manager screen.

The following two records are used to monitor the disk resources. For details about each record, see 5. Records.

PI_HPDI record

This record is used to obtain the number of I/O interrupts on the HBA installed on the host machine as performance data. This enables you to monitor the load on the physical HBA.
PI_VPDI record

This record is used to obtain the number of I/O interrupts on the entire virtual HBA allocated to LPARs as performance data. By monitoring the number of I/O interrupts per LPAR, you can monitor the disk access load of LPARs.

The following figure shows the range of performance data collected in each record.

Figure 1‒101: Correspondence between records and data collection ranges

To Page Top

(2) Monitoring examples

Using monitoring of the number of I/O interrupts of the physical HBA on the host machine running the virtual environment and that of the virtual HBA for LPARs as an example, this subsection explains the factors that cause insufficient disk resources and how to solve this problem. The following figure shows the items monitored here and the flow of actions to take.

Figure 1‒102: Monitored items and flow of actions

(a) Example of monitoring the number of I/O interrupts of the physical HBA on the host machine

You can monitor the number of I/O interrupts of the physical HBA on the host machine in the Requests field of the PI_HPDI record.

The following figure shows an example of monitoring the number of I/O interrupts of the physical HBA on the host machine.

Figure 1‒103: Example of monitoring the number of I/O interrupts of the physical HBA on the host machine

Monitoring template report to be checked

Host Disk I/O^#

#: The Host Disk I/O report does not display the Requests field in a graph. To monitor the number of I/O interrupts of the physical HBA on the host machine, customize the report definition before operation. For details about how to customize the definition, see (3) Customizing monitoring template reports.

In this example, HBA1 has more I/O interrupts, which means interrupts occur often and continually, and some measures must be taken.

(b) Example of monitoring the number of I/O interrupts of the virtual HBA for LPARs

You can monitor the number of I/O interrupts of the virtual HBA for LPARs in the Requests field of the PI_VPDI record.

The following figure shows an example of monitoring the number of I/O interrupts of the virtual HBA for LPARs.

Figure 1‒104: Example of monitoring the number of I/O interrupts of the virtual HBA for LPARs

Monitoring template report to be checked

VM Disk I/O^#

#: The VM Disk I/O report does not display the Requests field in a graph. To monitor the number of I/O interrupts of the virtual HBA for LPARs, customize the report definition before operation. For details about how to customize the definition, see (3) Customizing monitoring template reports.

In this example, LPAR1 has more continual interrupts and some measures need to be taken.

(c) Example of monitoring insufficient disk space for LPARs

To monitor whether disk space for LPARs is insufficient, use monitoring templates from PFM - Agent for Platform or PFM - RM for Platform. By using PFM - Agent for Platform or PFM - RM for Platform to monitor the items shown below, you can monitor what percentage of disks for LPARs are not used and identify which LPAR is running out of disk space.

For PFM - Agent for Platform

Monitoring template report to be checked: Free Megabytes - Logical Drive Status
Monitoring template alarm to be checked: Disk Space

For PFM - RM for Platform

Monitoring template report to be checked: Free Megabytes - Logical Disk
Monitoring template alarm to be checked: Disk Space

To Page Top

(3) Customizing monitoring template reports

This subsection explains how to show the Requests field in a graph by customizing the Host Disk I/O and VM Disk I/O reports in the monitoring template.

(a) Copying the underlying report

Before customizing the reports, you must copy the underlying report to the User Reports folder.

To copy the report:

In the navigation frame of PFM - Web Console, select the Reports tab.
In the Reports window, click System Reports and then RM VirtualMachine to expand.
Select the Host Disk I/O report or VM Disk I/O report that is stored in a desired folder.
Copy the selected report to the User Reports folder.

(b) Customizing the report

To customize the copied report:

In the navigation frame of PFM - Web Console, select the Reports tab.
In the Reports window, select User Reports and then select the Host Disk I/O report or VM Disk I/O report.
In the method frame, select the Edit method.

Click the Next button until the Edit > Components window appears.
Clear the check box under Graph for the Speed field, and then select the check box under Graph for the Requests field.

Click the Next button to show the Edit > Graph window.
Change the value of Y-axis in Axis labels from Speed[Kbytes/Sec] to Requests.
Click the Finish button.

To Page Top