Tech TopX: VM Right Sizing & Behavior Learning

Tech TopX: VM Right Sizing & Behavior Learning


Prism Central 5.5 comes with behavior
learning tools that use its machine learning capabilities to analyze users
consumption over a period of time, and provide tools to monitor resource usage,
identify any abnormal behavior, and also guide resource planning. Today, we will
discuss one of these tools: VM right-sizing. Welcome back to another episode of Tech TopX, my name is Vineeth and I’m a systems
reliability engineer with Nutanix. In order to make the best use of
virtualization’s most promising feature – which is to reclaim vast amounts of
resources that stood forever idle on physical servers – the biggest challenge
an administrator faces today is to size the VMs in an efficient manner, so that
no VMs run out of resources and that no resources are wasted. When we analyze
performance issues in a cluster, or even evaluate future resource needs, it can be
useful to identify VMs that are not optimally sized. The new VM right sizing
feature identifies inefficient VMs such as ones that consume too many resources,
are constrained, over provisioned, or are inactive. There are many ways to view the
VM efficiency data on Prism Central. The VM efficiency widget exists on the home
dashboard by default, however if it doesn’t it can be easily added using the
add widget button on the top right. Let’s click on view all inefficient VMs. Here
you can see a list of inefficient VMs across all the clusters that are
registered to the Prism Central. The efficiency column shows which category
of inefficient VMs does a VM fall under. The efficiency detail column shows why a
VM is classified as inefficient. As you can see, there are VMs that are
considered as constrained and over provisioned. Let’s look at one such VM.
Click on the VM name and go to the metrics tab. Here you can see that the
CPU usage classifies the VM as over-provisioned as the CPU consumption
has constantly been under the lower threshold of 20% in the past 21 days. Whereas the memory usage
classifies the VM as constrained since its memory consumption has been
constantly exceeding the upper threshold of 95%. In order to see VMs on a particular cluster you can either use
the filters and choose the cluster, or you can simply sort it by the cluster
name. You can also go to the explore tab, clusters, click on any cluster, go to
capacity, and then click optimize resources to view all the inefficient
VMs. By sizing the VMs right you will have better performance, happy users, and less support calls at 3:00 a.m. I hope this video was helpful.
Please check out our channel for more videos on AOS 5.5 and thank you
for watching.

Leave a Reply

Your email address will not be published. Required fields are marked *