Overview of VMWare HA - High Availability

<

Overview of VMWare HA - High Availability

Written by Damian Murdoch   

Wednesday, 28 June 2006

VMWare High Availability or HA is another Virtual Infrastructure 3 component that adds tremendous value to the product line. This product in earlier versions was known as DAS or Distributed Availability Services. High availaility allows Virtual Machines to failover automatically between ESX hosts in the same cluster. This article will give an overview of the product and help you decide on the value it can add to your business.

VMWare HA is something that we have been waiting for now since the original release of ESX. The High Availability component leverages the technology of Virtual Center, ESX server and shared storage to provide value to the business. It provides this value by maximising service uptime. The administrator has many options for configuration and management of HA. right down to customising behaviour for each of the virtual machines.

While the high availability option was available before if you were a guru and could do some nice scripting and cron jobs, it was not for the average administrator. The reason behind this was the design of the virtual machine configuration in earlier versions of ESX server. The configuration file associated with the VM's, the .vmx file was stored individually on the host it ran on. In Virtual Infrastructure 3 it now lives on shared storage. This enables multiple hosts to access that file should the host it is running on crash or go down.

HA is configurable through Virtual Centre 2.0 and provides many options for config. To configure HA you just need to specify a new cluster of machines. This configuration wizard will run you through options such as the number of host failures allowed and whether you want to start the vm's if the resources are tight. You have the options of setting high priority VM's that provide critical services to start before others that provide less critical services. Another configurable option is what to do if a Virtual Machine is identified as isolated, you have the option to turn it off or leave it on.

The HA module actually uses an engine that VMWare has licensed from Legato called the AAM. Virtual Center integrates with this engine to report the state of the Virtual Machines. The DRS component discussed in a previous article is used for analysis the hosts for the VM restart. DRS can then be used to rebalance the load after HA restarts the machines. You can use the standard alarm settings in VMWare to alert you if a host crashes or a bunch of VM's are migrated between hosts if you like.

Keep in mind that HA works within a cluster. You cannot use HA between clusters or datacenters. So this is not your DR solution but it is a solution for priority services that require high levels of uptime. This way you can at least meet those crazy sla's!

ZT: