Solved

Nutanix equivalent to VMware's vCenter HA for management high availability?

  • 10 March 2022
  • 3 replies
  • 177 views
D

New to Nutanix and trying to understand HA for management like in VMware.  In VMware we setup vCenter HA with an active/passive/witness.  Is there something similar in Nutanix?

icon

Best answer by Jason Hendren 15 March 2022, 15:22

View original
This topic has been closed for comments

3 replies

J
Rank
Userlevel 2
Badge +1

​

Hi @danfrnot,

In Nutanix, Prism's cluster management service runs on every CVM with an elected Prism Leader responsible for handling HTTP requests. If the Prism Leader fails, the cluster will automatically select a new one. When a CVM that is not the Prism Leader gets an HTTP request, it will permanently redirect it to the current Prism Leader.

Here's an excellent resource link for understanding more about Nutanix's Prism architecture .

Let me know if that helps answer your question.

Also, check out KB-12730 for a comprehensive list of resources for new Nutanix family members! 

​

Moustafa Hindawi
Userlevel 5
Badge +6

Hi @danfrnot , 

As Jason ask the prism element instance runs on every CVM and one elected as leader, you connect to Prism Element by virtual IP and the virtual IP redirect you to the leader, when a CVM which host the leader failed, the cluster elect new leader.

Moustafa Hindawi | Blogger | Nutanix Technology Champion | NCSE | NCM-MCI | VMCE | 2xVCAP-DTM & DCV | 2xVCP-NV & CMA | LinkedIn: https://www.linkedin.com/in/moustafaadelmoustafa/ | My Blog: https://cloudwhales.co.uk/author/mhindawi/
DavidN
Userlevel 3
Badge +6

I second Jason and Moustafa’s notes/responses but would like to add some thoughts:

I think your question is really multiple questions rolled into one, so it’s best to address each one with details.

  1. How does the cluster deal with the loss of a node with respect to cluster services?
  2. How does the cluster handle resources during a node failure?
  3. How does a cluster handle running VMs that were on the failed node?

For question 1: some background with respect to the various services that make up the cluster and how they handle recovery - along the line how Jason’s reply was pointing 
https://www.nutanixbible.com/2f-book-of-basics-cluster-components.html

Notice that many of these services use a multi-node model - i.e. in general most services either run on every node and elect a leader OR they run on at least 3 nodes and elect a leader (that can failover if it stops responding)

 

For questions 2 & 3: there is an explicit checkbox to have the cluster ‘reserve’ enough resources so that a node’s failure won’t bring things down...

https://portal.nutanix.com/page/documents/details?targetId=Web-Console-Guide-Prism-v5_20:ahv-acropolis-cluster-setting-ha-t.html
 

Another excellent KB goes more into details on AHV HA

https://portal.nutanix.com/page/documents/kbs/details?targetId=kA00e000000LIQUCA4

 

And other posts on this topic:

https://next.nutanix.com/archive-44/episode-1-acropolis-hypervisor-vm-ha-part-1-5595

https://next.nutanix.com/archive-44/episode-1-acropolis-hypervisor-vm-ha-part-2-5597

 

Hope this helps.

Terms & Conditions