VM Resource Distribution

Sam Kirchoff

VM resource distribution is the first in a series of graphs designed to help understand allocation of resources at the VM level in virtualized environments.

All Virtual Machines have some allocation of vCPU, Memory, and Capacity. However, they are not all allocated equally.

In many environments only a smaller percent might represent disproportionate ownership of these resources while the others are rather standard.

This graph tracks a 95th % value for the following:

  • 95th % value of the total number of assigned vCPUs to all VMs.
  • 95th % value of the total assigned memory in GBs to all VMs.
  • 95th % value of assigned capacity to all VMs.  

Unique VMs

If any Virtual Machine exceeds the allocation of 95th % for any of these resources, it is considered unique.

These VMs are most likely consuming the most cost and would be the least likely to perform well in a Public Cloud.

It would be suggested that you run a Live Optics Server/Virtualization assessment and directly target these VMs to get further detail as to their workloads, configuration and performance.

Standard VMs

Any Virtual Machine whose resource allocation in all three categories is below the 95th % will be considered standard.

The graph itself plots the number of Standard vs. Unique VMs. It will also summarize the total allocation of resources for both standard and Unique VMs.

Each environment is different

Just as no business is the same, there is no two environments that are identical. In some, the resource allocation might be very uniform. This would lead to largely all blue or all standard VMs.

However, in others there might be any number of VMs that exceed the values to be considered Unique.

In general terms, these Unique VMs might have workloads that are not generic enough to be considered for Cloud or Hyperconverged.

As an example, VMs that are disproportionately heavy on storage assignments might be better suited for a more traditional infrastructure where CPU and Capacity can scale independently.

A second example might be those VMs that have a very high numbers of assigned vCPUs.

Public Cloud providers have a limit on the number of vCPUs that can be assigned to a Virtual Machine. If the VM really needs this number of vCPUs to accommodate the workload, it might not be well suited for a Public Cloud.

Example Graph

In the graph below we can see that only 4 of the 28 VMs exceed this 95th % limit in any of the three categories.


However, a heavy amount of resources is assigned to just to these 4 VMs.

  • 24% of all vCPUs are assigned to these VMs.
  • 29% of all memory is assigned to these VMs.
  • 39% of all allocated capacity is assigned to these VMs.

Note: Only 22% of all used capacity resides on these 4 VMs. This is an indicator that these 4 VMs are possible overprovisioned in capacity

Was this article helpful?
3 out of 3 found this helpful
Please sign in to leave a comment.