Episode 1: Acropolis Hypervisor: VM HA Part 1

  • 13 October 2015
  • 0 replies
Episode 1: Acropolis Hypervisor: VM HA Part 1
Userlevel 7
Badge +35
This post was authored by Manish Lohani, Director of Product Management

Acropolis 4.5, which became generally available last week, introduces some important virtualization management features as we rapidly continue to build our solution. VM high availability (VM-HA) is one such popular and important feature provided by the virtualization management layer - and we are providing this enterprise class functionality that is consumer grade from ease of use standpoint.

So, what makes VM-HA easy to use? Out of the box, Nutanix cluster with AHV is pre-configured to provide “Best Effort High Availability” and will respond to node failures by restarting VMs on healthy nodes as long as the cluster has available capacity.So, if your workloads on the cluster do not warrant reserving spare capacity for failover, then no configuration is needed. However, to reserve spare failover capacity, you can check “Enable HA” option in the Prism UI any time after a cluster is created. That’s all the configuration you need to use VM HA for AHV !

You may be thinking - “Wait ! That’s it? Where did I tell VM-HA what admission control policy to use, how many node failures to protect against, what should be the host isolation response etc. etc. - All the usual configuration one has to perform when enabling a feature like high availability?”

These are great questions! Nutanix has implemented this feature in such a way that software minimizes the need for you to configure it and does “the right thing” whenever it can. The software is sophisticated to not only deal with all the complexity under the hood but also to simplify it for the user. However, while preserving simplicity, VM-HA still has the necessary controls available for advanced users who want them or may need them. Users can still enable / disable VM-HA per VM, specify relative priorities to restart VMs in case of a node failure, manually override auto-selected admission control policies and specify the number of host failures to tolerate.

Here, I would like to gently remind the readers that VM-HA functionality is about high availability of VM instances on AHV and is different from high availability of data (virtual disks) on a Nutanix cluster or high availability of Nutanix controller VM (CVM). Data availability is achieved by storing multiple (2 or 3 as per configured replication factor) copies of data across nodes within a cluster and has been available across multiple hypervisors since early days of the Nutanix product along with Nutanix controller VM (CVM) high availability.

To learn more about the details of VM-HA functionality, architecture and other AHV management features in 4.5, stay tuned for my next posts in this series.

Meanwhile, if you just set up your first AHV based cluster - and have not set up VM-HA yet, don’t stress, it will only take a minute. And then time for a break !

0 replies

Be the first to reply!