Increasing memory and CPU resource allocations

Virtual SMU Installation and Upgrade Guide for Hitachi NAS Platform

Version
15.1.x
Audience
anonymous
Part Number
MK-92HNAS074-20

Before you add extra managed servers or clusters to the virtual SMU, increase the memory and the CPU resource allocations to reserve sufficient resources for each VM.

Note: The exact user interface may differ from the screenshots shown below, depending on the version of the vSphere client in use.

The following example is for four managed servers.

  1. Power off the VM.
  2. In the vSphere Client, right-click the VM and select Edit Settings to open the Virtual Machine properties dialog box.
  3. Under Virtual Hardware, select CPU:
    1. Either set the CPU option to four or set the Cores per Socket option to two to make a total of four sockets.
    2. (Optional) Although CPU reservation is not required, you should increase the CPU reservation if the host supports it.
    3. (Optional) If other VMs on the host can starve the virtual SMU of resources, you can set Shares for CPU (and Hard Disk) to High. This prioritizes the virtual SMU over VMs with a Normal or Low setting.
    Virtual Hardware: CPU
  4. Under Virtual Hardware, select Memory:
    1. Increase the memory value to 4GB.
    2. (Optional Best Practice*) Set the reservation to 4096GB.
    3. (Optional) If other VMs on the host can starve the virtual SMU of resources, you can set Shares for Memory to High.
    Virtual Hardware: Memory
  5. Click OK to save your changes, and then close the dialog box.
  6. Right-click the VM and select Edit Settings again to verify that your memory and CPU settings are correct.

*Best Practice

Although resource reservations are not required, it is best to reserve a portion of the host's physical RAM to guarantee the responsiveness of the virtual SMU and its quorum device. Ultimately, if you oversubscribe the host, do so responsibly to ensure that the virtual SMU is not starved of resources.

The critical virtual SMU requirement is that the quorum device must respond to cluster heartbeats (over UDP) within five seconds to prevent the possibility of dependent and degraded HNAS clusters rebooting. Resource reservations are just one way of achieving this requirement. VMware also provides other mechanisms to ensure VM responsiveness and to protect against resource starvation.