1.4.5 Monitoring disk resources

This subsection explains how to monitor the disk resources of a VMware system.

Organization of this subsection

(1) Overview
(2) Monitoring examples
(3) Other monitoring examples

(1) Overview

With a VMware system, you can create a datastore on a physical disk and share it with multiple virtual machines. A virtual disk created on the datastore is allocated to each virtual machine. The OS running on each virtual machine recognizes the allocated virtual disk as a normal physical disk and uses it as logically partitioned logical disks.

The following two types of disk resources are available:

Disk I/O resource

This is a resource related to disk access statuses.
Disk space resource

This is a resource related to disk space.

Since the physical disk is shared in a virtual environment, it is accessed by the individual virtual machines simultaneously. If multiple disk commands are issued simultaneously, the disk I/O resource is distributed according to share ratios that have been set up by the virtual environment software.

Disk commands from a virtual machine for which a large share ratio is set are executed at higher priority. Consequently, disk commands from a virtual machine with a small share ratio may be discarded without being executed. Discarded disk commands are re-executed after a time.

If the rate at which commands are discarded becomes high, disk access is delayed, adversely affecting the performance of the virtual machine. By monitoring the performance data of the disk I/O resource, you can detect such performance deterioration of the virtual machine, and you can thus take the necessary corrective action.

Furthermore, you can check whether there is a disk space insufficiency by monitoring the performance data of the disk space resource.

The six records described below are used to monitor the disk resource. For details about records, see 5. Records.

PI_HPDI record

This record is used to monitor the performance data of the physical disk. It shows the disk I/O resource as viewed from the physical server.
PI_VPDI record

This record is used to monitor the performance data of the physical disk being used by the virtual machine. It shows the disk I/O resource as viewed from the virtual machine.
PI_HLDI record

This record is used to monitor the performance data of the physical server's datastore. It shows the datastore space resource as viewed from the physical server.
PI_VLDI record

This record is used to monitor the performance data of the guest OS's logical disk. It shows the disk space resource as viewed from the guest OS.
PI_VVDI record

This record is used to monitor performance data of the virtual disk used by the virtual machine. It shows the virtual disk I/O resource as viewed from the virtual machine.
PD_VDKD record

This record is used to monitor performance data of the virtual disk used by the virtual machine. It shows the allocated virtual disk space resource as viewed from the physical server.

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

Figure 1‒22: Correspondence between records and data collection ranges

Important

The free space of the VMware datastore is not updated regularly by default. Use the Last Update field of the PI_HLDI record to make sure that the last update time of the VMware datastore is periodically updated.

For details, see 2.5.1(5) Confirmation of Host Logical Disk Status (PI_HLDI) record.

To Page Top

(2) Monitoring examples

Using monitoring of disk resources for vhost1 and vhost2 running on a physical server called host1 as an example, this subsection explains the problems that might occur in the disk resources and how to solve them. The following figure shows the items monitored here, and the flow of actions to take.

Figure 1‒23: Monitored items and flow of actions

(a) Example of monitoring the disk command discarding rate of a physical disk being used by a virtual machine

You can monitor the disk command discarding rate of a physical disk being used by a virtual machine in the Abort Commands % field of the PI_VPDI record. Note that you can monitor this item with an alarm provided in a monitoring template.

The following figure shows an example of monitoring the disk command discarding rate of a physical disk.

Figure 1‒24: Example of monitoring the disk command discarding rate

Monitoring template report to be checked: VM Disk Abort Commands
Monitoring template alarm to be checked: VM Disk Abort Cmds

If there is a virtual machine with a high disk command discarding rate, adjust its share of the physical disk. In this example, the disk command discarding rate for the physical disk being used by vhost2 is high, and increasing vhost2's share can solve this problem. For details about how to adjust the share, see the documentation for the virtual environment software.

(b) Example of monitoring the space usage of a physical server's logical disk

You can check whether a physical server's logical disk has sufficient free space based on the space usage. Note that you can monitor this item with an alarm provided in a monitoring template.

You can check the space usage in the Used % field of the PI_HLDI record.

The following figure shows an example of monitoring the space usage of a physical server's logical disk.

Figure 1‒25: Example of monitoring the space usage of a physical server's logical disk

Monitoring template report to be checked: Host Disk Used
Monitoring template alarm to be checked: Host Disk Usage

In this example, the space usage of Disk 2 is high. If the space usage is high, you can solve this problem by expanding the logical disk space through the addition of a physical disk. For details about how to expand the disk space, see the documentation for the virtual environment software.

(c) Example of monitoring the space usage of a virtual machine's logical disk

You can check whether a virtual machine's logical disk has sufficient free space based on the space usage. You can check the space usage in the Used % field of the PI_VLDI record. Note that you can monitor this item with an alarm provided in a monitoring template.

The following figure shows an example of monitoring the space usage of a virtual machine's logical disk.

Figure 1‒26: Example of monitoring the space usage of a virtual machine's logical disk

Monitoring template report to be checked: VM Disk Used
Monitoring template alarm to be checked: VM Disk Usage

In this example, the space usage of Disk 2 being utilized by vhost1 is high. If the space usage is high, you can solve this problem by expanding the logical disk space allocated to the virtual machine. For details about how to expand the disk space, see the documentation for the virtual environment software.

To Page Top

(3) Other monitoring examples

Monitoring examples follow that use monitoring templates other than those explained in (2) Monitoring examples.

(a) Report that displays the disk I/O state of a physical server

Figure 1‒27: Example of monitoring a physical server's disk I/O state

Monitoring template report to be checked: Host Disk I/O

(b) Report that displays the disk I/O state of a virtual machine

Figure 1‒28: Example of monitoring a virtual machine's disk I/O state

Monitoring template report to be checked: VM Disk I/O

(c) Report that displays the usage status of a physical server's logical disk

Figure 1‒29: Example of monitoring the usage status of a physical server's logical disk

Monitoring template report to be checked: Host Disk Used Status

(d) Report that displays the usage status of a virtual machine's logical disk

Figure 1‒30: Example of monitoring the usage status of a virtual machine's logical disk

Monitoring template report to be checked: VM Disk Used Status

(e) Report that displays the capacity of the virtual disk allocated to a virtual machine

Figure 1‒31: Example of monitoring the capacity of the virtual disk allocated to a virtual machine

Monitoring template report to be checked: VM Virtual Disk Allocation Value

By outputting this report to a CSV file and summing up the latest values, you can identify the capacity of the virtual disk allocated to a datastore.

When a datastore is shared by multiple physical servers, information regarding each one of these physical servers is acquired and data belonging to the same Datastore ID or Datastore Name field is summed up. This allows you to identify the capacity of the virtual disk allocated across the entire system.

Furthermore, if ESXi 6.5 or later is monitored, you can identify the virtual disk shared by multiple virtual machines by using the Disk UUID field.

To Page Top