VMHA

·      Let’s say that you have an ESX server hardware failure (say a bad CPU or even disconnected network cable)

·      Or, you could have an ESX server software OS crash (unlikely)

·      VMHA powers all VMs running on that server on other servers in the VMHA cluster and Apps are up in the time it takes guest OS to boot

·      Downtime=Restart Time

VMHA Saving the Day

·    VMHA monitors not only ESX host failures but also Guest OS failures

·     Uniform HA protection for all VM guests and all applications, no matter the OS or app

·     Smart failover to best ESX host (requires DRS)

·     Supports up to 32 ESX servers in a cluster

·     Enhanced isolation response

·      Heartbeats are default of 1 second with 15 seconds to assume an ESX host is dead

·      Failed ESX server will reboot in maintenance mode

·      Admission Control ensures that resources are available (could stop power on of a VM, migrate a VM or increasing CPU/RAM reservations

Requirements for VMHA

·     Shared storage for VMs running in HA cluster

·     Can use DRS with VMHA or just VMHA only

·     vSphere Essentials Plus, Standard, Advanced, Enterprise, Enterprise Plus

·     At least two Hosts in the cluster

·     Create a VMHA enabled cluster

Configuring VMHA, step by step

·      Have all VMs on shared storage between ESX servers

·      Create a cluster and enable VMHA □Add ESX hosts to the cluster □Optional Configuration:

·      Cluster rules

·      Guest VM restart priorities

·      Guest VM Monitoring of OS

Best Practices for VMHA

·    Keep an eye on Cluster Validity

·    Recommend you disable host monitoring as you make changes to your network or dvSwitches

·    All networks and VMs on HA clusters must have compatible networks

·     By default network isolation IP is the default gateway but you can configure others

·     Use network redundancy between ESX servers

·      Admission control ensures that recourses are available (Could Stop Power on of a VM or Increasing CPU/RAM Reservations)

VMHA Master/ Slave

·      Master Host per cluster known as the Fault Domain Manager Master

·      FDMS host is determined via an election process

·      FDMS host with the most mounted database has better election chances.

·      VMHA check for host failure using the management network and via data store heartbeat.

What is the task assigned to FDMS Host

·      Monitors the state of Slave hosts

·      Monitors the state of all protected VMs.

·      Manages the list of cluster hosts and protected VMs.

·      FDMS is vCenter management interface in to the clusters health state.

Auto Load Balancing with DRS

·   Hosts and Clusters provide the resources

·   The VMs consume the resources

·   Goals of DRS-

ü     Prevent one VM from monopolizing all resources

ü     Guarantee service levels

ü        Offer most efficient use of server hardware

ü     Make your life as a VMware admin easier

Auto Load Balancing with DRS

·       DRS is enabled on a cluster

·      Automation levels are:

1.      Manual - suggestions

2.      Partially automated - auto place on power on & suggestions

3.   Fully automated - you set the migration threshold

Ø  You can create migration rules to keep VMs together or apart

Ø  DRS can auto-place new VMs in the cluster

Requirements to use DRS

·     vSphere Enterprise or Enterprise Plus

·     Shared storage between hosts

·    All VMs in the cluster must be on that shared storage

·      DRS will use VMotion so it needs to work between hosts (beware of CPU compatibility issues & if so, checkout Enhanced VMotion Compatibility / EVC)

·      Checkout the DRS tab and Resource Allocation tab on the cluster as well as the cluster properties.