chris.graham

Turbo Tip: Pinning Nutanix Controller VMs (CVMs) in Non-Standalone Mode

Blog Post created by chris.graham Expert on Dec 16, 2015

This article is intended for users of Turbonomic NOT RUNNING Acropolis Hypervisor (AHV). If you are running Acropolis Hypervisor, please stop reading.

 

Seriously stop reading.

 

Turbonomic integrates with Nutanix Xtreme Computing Platform (XCP) to assure application performance in homogenous Nutanix environments and heterogeneous environments alike. Specifically, Turbonomic pulls Nutanix provider entities into its Supply Chain using the vCenter/Hyper-V APIs, the Nutanix Prism API, or both. In so doing, Turbonomic understands the unique resource demand profiles of every XCP workload: compute, SSD storage, HDD storage, and network, and drives the specific sizing and placement actions that deliver the best possible performance.

 

When you specify only Nutanix as a storage target and are using Acropolis Hypervisor as the underlying hypervisor, you are running in Standalone Mode. When you specify both Nutanix as a storage target and also vCenter or Hyper-V as the underlying hypervisor, you are running in Non-Standalone Mode. The steps in this article are only necessary if you are running in Non-Standalone Mode.

 

Nutanix architecture is unique in that it (a) virtualizes the storage controller and (b) Distributes writing of the data path to multiple nodes across N+X virtual storage controllers (referred to as the Resiliency Factor: RF1, RF2, RF3, etc.). Controller VMs contain all of the software that drive the unique behavior of Nutanix appliances.

 

Controller VMs are statically placed on each Nutanix node, tethered by the PCI card. However, in Non-Standalone Mode, Turbonomic is capable of discovering Nutanix CVMs and will attempt to migrate them if there is a better place within the federation to run.


This article walks through the 2-step process of discovering and pinning Nutanix CVMs such that Turbonomic does not migrate them:

  1. Creating a dynamic custom group of Nutanix CVMs
  2. Disabling migration of these CVMs by Turbonomic

 

STEP ONE: CREATING A DYNAMIC CUSTOM GROUP OF NUTANIX CVMs*

 

By default, Nutanix CVMs generally follow the naming convention NTNX-(Serial Number)-A-CVM . Since Turbonomic Dynamic Custom Groups are defined using regular expression, we can easily define the group of VMs following this convention.

  • Within Turbonomic, go to Policy > Group Management and clickScreen Shot 2015-12-16 at 4.53.08 PM.png
  • In the Group Builder modal, under the Properties tab, name the group "NutanixCVM" (or something descriptive to your liking)
  • Under the Members tab, from the "Search for:" dropdown, select "Virtual Machines" and "by name"
  • Under "VirtualMachine with a name matching the regular expression", select "equals" and type NTNX.*CVM
    • This Regex will search globally to find all virtual machines matching the NTNX-(Serial Number)-A-CVM convention.
  • Click "Find Matches" and then "Create Group"

Screen Shot 2015-12-16 at 4.58.45 PM.png

 

STEP TWO: ENFORCE A WORKLOAD PLACEMENT POLICY ON THE NutanixCVM GROUP

 

 

  • Within Turbonomic, go to Policy > Action > VM
  • In the Scope: Group column, go to the "My Groups" folder and select the NutanixCVM group you just created
  • In the Parameter column, under Action Mode Settings, select "Disabled" from the dropdown adjacent to "Move"
  • Click "Apply Settings Change"

^62AF6BAF44E48AC32ED870544788D74B0BC0533D154E94BD9B^pimgpsh_fullsize_distr.jpg

 

Your Nutanix CVMs are now pinned to their nodes within Turbonomic.

 

*This article assumes you maintain the default naming convention of Nutanix CVMs. If you have edited the naming of CVMs, simply modify the Regex search in Step One to reflect this change.

 

Thanks to discoposse for his help on this!

erichorsman for your reference!

Outcomes