1.4.4 Monitoring the memory resource
This subsection explains how to monitor the memory resource of a VMware system.
- Organization of this subsection
(1) Overview
In a VMware system, multiple virtual machines share the memory of the physical server. The memory resource allocated to each virtual machine is called virtual memory. The OS running on a virtual machine recognizes virtual memory as normal physical memory.
The memory resource of the physical server is distributed according to each virtual machine's memory resource requirement. However, if the combined total of the memory resource requirements of the individual virtual machines exceeds the memory resources on the physical server, the required amount of memory resources cannot be distributed, resulting in a virtual memory resource shortage. This adversely affects the performance of the virtual machine. By monitoring the memory performance data, you can detect such performance deterioration of the virtual machine.
Furthermore, in the virtual environment, swapping is used. Swapping means using some disk space as memory. Disk space that can be swapped is called a swap. Through swapping, you can use a memory resource amount that is larger than the amount of memory installed in the physical server.
Since disk access speed is slower than physical memory, using a swap adversely affects the performance of the virtual machine. Therefore, we recommend that when you monitor memory resources, you also check the swapping status.
The following two records are used to monitor memory resources. For details about records, see 5. Records.
-
PI_VMI record
This record is used to monitor each virtual machine's memory usage and the number of external and internal swaps used.
-
PI_HMI record
This record is used to monitor the usage of physical memory by the virtual machine monitor and by each virtual machine, and the status of internal and external swaps used by the entire physical server.
The following figure shows the range of performance data collected in each record.
|
(2) Monitoring examples
Using monitoring of a physical server on which the virtual environment is running as an example, this subsection explains the factors that cause insufficient memory resources and how to solve these problems. The following figure shows the items monitored here and the flow of actions to take.
|
(a) Example of monitoring the total memory usage of a physical server
You can monitor the total memory usage of a physical server in the Total Used % field of the PI_HMI record. The total memory usage indicates the usage of all memory resources (physical memory resources, internally swapped resources, and externally swapped resources) provided on the physical server. If this value is large, the memory resources of the physical server are considered insufficient.
The figure below shows an example of monitoring the total memory usage of a physical server.
|
- Monitoring template report to be checked
- Monitoring template alarm to be checked
In this example, the value of Total Used %, which indicates the combined total of Used %, VM Swap Used %, and Host Swap Used %, exceeds 100%. Therefore, the memory resources of the physical server are considered insufficient.
In this case, reassess the configuration information of the virtual machine. If the total memory usage value does not improve even after the configuration information has been reassessed, consider either adding memory resources to the physical server or adding a physical server.
(3) Other monitoring examples
(a) Report for evaluating the memory usage of a physical server
|
- Monitoring template report to be checked
(b) Report for evaluating the memory usage of a virtual machine
|
- Compound report to be checked (see 1.10)
-
Monitoring of the upper limit of memory allocation in a virtual machine