Logical Segregate production and Non-prod VM in a HCI cluster

  • 8 April 2020
  • 3 replies

We have many clusters in HCI . Now we want to locally segregate prod VM n NOn -prod VM in a cluster . how can we do ? Do we have concept of resource groups?


Best answer by AntonioG 25 April 2020, 01:48

View original

This topic has been closed for comments

3 replies

Userlevel 3
Badge +4

Hi @Anand_Saw 

Prism Central is structured for policy-based management of VMs based on assigned categories. I think this would be the functional way of doing what you describe here. As an example, these categories can be used to assign user permissions, or configure entity-centric backups which apply for the whole category.



Prism Element does not include this feature, so if you want to be able to quickly distinguish prod vs non-prod in PE you might look at naming conventions or annotation on the VMs which could then be used in search to narrow down the VMs displayed.

Thanks Jeremy ,

I was not only looking to categories Prod and Test VMs but to partition them so that compute and storage are not linked . This would protect my Mission critical server from latency or storage issues . 

Since I am using Esxi as hypervisor . I guess I have only viable option to create Resource Pool .

Any other option ?


Userlevel 2
Badge +4

Hi @Anand_Saw 

I got a couple of opinions from the team, hope they help:

1) The Best Practice for this scenario, for an Enterprise enviroment is to have a separate DEV (Test) Cluster.For those scenarios where you are  unable to following an appropriate data center design, or common sense. One common way to do this in an vSphere enviroment would be to have two separate vSphere Clusters .Cluster A 3-Node with Production VMs, Cluster B 1-Node or more for Test/Dev VMs. In this scenario all 4+ nodes would be part of the same Nutanix Cluster. They have to be aware that vSphere HA may be configured to support 1 node failure in each vSphere Cluster, however, only 1 node failure would be survivable in the Nutanix Cluster.

2)One other solution is to create a rule in vCenter saying that all DEV/TEST VMs can run only in one or 2 particular hosts. So if those hosts become constraint,  critical VMs will be migrated by DRS but dev/test VMs will remain there.