vSphere 5 HA features

In vSphere 5.0, the vSphere HA feature has been completely rewritten by VMware from the ground up. This was done to increase the scalability, reliability, and usability of vSphere HA

Changes in VMware HA :
Scalability :-
One of the biggest changes with vSphere HA is that the concept of primary and secondary nodes has been
completely removed. The new model incorporates a master-slave relationship between the nodes in a cluster,
where one node is elected to be a master and the rest are slaves. The master node coordinates all availability
actions with the other nodes and is responsible for communicating that state to the VMware vCenter Server.

Realibility :-
One enhancement was the elimination of dependencies on any external component by vSphere HA. Specifically,vSphere HA no longer has any type of dependency upon DNS resolution by each host in the cluster. Eliminating this reduces the odds that an outage of an external component will have an effect on the operation of vSphere HA.

Another enhancement is the ability to enable communication between the nodes within a cluster through the
storage subsystem. Now vSphere HA will utilize multiple paths of communication, through the network and
storage. Not only does this allow for a greater level of redundancy, but it also enables better identification of the health of a node and the virtual machines running on it.

Virtual Machine Scalability and Added Device Support :-
Up to 32 virtual CPUs (vCPUs) and up to 1TB of RAM
Enhanced graphics capabilities, including 3D graphics support that enables a richer desktop experience
Broader device coverage, including support for 3.0 USB devices, smart card readers and EFI BIOS
New user interface support for configuring multicore vCPUs
Support for new guest operating systems including Mac OS X Server v10.6 (“Snow Leopard”)


Compatibility with Older Versions of VMware Tools and Virtual Hardware


vSphere 5 HA replaced AAM with FDM (Fault Domain Manager)
If you are doing host maintance or network uncheck host monitoring in the HA
Master-Slave /etc/opt/vmware/fdm

Network Partition: is the term used to describe the suitation in which one or more slave hosts cannot comminicate with the Master even though they still in network connected .HA uses heart beat datastores to detect wheather the partitioned hosts are still live and whether actions needs to be taken to protect VM's on those hosts.


Network Isolation :
Is the situation in which one or more slave hosts have lost all managment network connectivity.Isolated hosts can neither communicate with HA master or commnicated with other ESXi hosts.