Using vROps to Right-size Hyper-V VMs

by Kyle Wassink on July 5, 2018

I was working with a customer recently who had both VMware ESXi and Microsoft Hyper-V environments that they needed to manage. It was clear that neither Hyper-Visor was going away, and there was an urgent need to simplify and combine the VM memory right-sizing process in one place.

Enter Blue Medora and vRealize Operations
This customer was already using VMware vRealize to manage much of their stack. With vRealize Operations, they could access VMware Hyper-Visor performance analytics and get memory right-sizing, out-of-the-box. Their issue and focus was on validating that Hyper-V could be right-sized—specifically, that Hyper-V VM memory could be optimized. We quickly added Blue Medora’s management pack for Hyper-V to the mix to provide Hyper-V metrics to vROps.

Hyper-V supports two memory configurations – dynamic and static. In VMware memory management terms, Hyper-V dynamic memory makes you assign a memory reservation (startup RAM in Hyper-V) and a memory limit (max RAM). For static memory, you provide a single RAM amount which acts as both the reservation and limit (i.e.:, it will always have guaranteed access to the configured amount of RAM).

The key data points
Digging into the memory metrics available at the Hyper-V VM level gave them all the information they needed to have a discussion with the application owner about reclaiming memory. The metrics below were the key ones:

For Static Memory Hyper-V VMs:

  • 30 day average demand (utilization)
  • 30 day max demand (utilization)
  • Startup memory

For Dynamic Memory Hyper-V VMs:

  • 30 day average and max demand (what the VM is requesting, since memory isn’t guaranteed)
  • 30 day average and max assigned (what the VM actual received)
  • Startup memory

Making the data usable
All the key information (plus some extra) was combined into a vROps List View, seen below.

Using vROps to Right-size Hyper-V VMs
Figure 1: Sample output in Blue Medora’s lab environment. Note that the sum of the max memory demand is under 4GB for the past 30 days, but over 9GB is provisioned as “Startup memory” (i.e., guaranteed memory)

In addition to the rightsizing metrics, some contextual information was also included such as the Hyper-V host the VM was on, what memory configuration it used, and whether vROps thought the VM was oversized using its capacity analysis.

With a report showing upwards of 50% oversizing, this team has a much better likelihood of making a case for reclaiming and repurposing those unused memory resources.

Automated report and capacity analysis — saving time, increasing efficiency
In vROps, Views are the building blocks of reports. To further simplify the rightsizing exercise, the View in Figure 1 could be put in a report format (PDF, CSV). It could also be setup to automatically run and be delivered to the manager’s email every month. Leveraging vROps and Blue Medora to do all the data collection, reporting, and delivery can shave a lot of time off your rightsizing efforts, as it now does for this Blue Medora client.

It’s also worth noting that the Hyper-V pack integrates with the vROps capacity engine. In the View we are using a simple true/false “is Oversized” output from this analysis, but there is an entire section of vROps focused on rightsizing. See the screenshot below for an example of what that looks like for a Hyper-V VM.

Using vROps to Right-size Hyper-V VMs
Figure 2: The Capacity Remaining analysis in vROps 6.6 for a Hyper-V VM. We can see for memory utilization (bottom), this Hyper-V VM’s average demand is 29.55% of its memory allocation, and vROps is recommending reclaiming 65.27% of the allocated memory.

For more information on Blue Medora’s vRealize Operations management pack for Hyper-V visit our website. And you can learn how to get started using it by reviewing this implementation blog post.

Get started

Take the quick 5-step assessment to see how BindPlane can meet your needs.

True Visibility
BindPlane for VMware vRealize Operations

True Visibility allows cloud management teams to use VMware vRealize’s powerful machine learning and capacity planning engine across their entire hybrid cloud environment.

BindPlane for the Data Layer

SelectStar makes it easier for DBAs and developers to monitor database performance during cloud transformation and DevOps adoption.