ShoreTel fail over options using Vmware – Part 3 the “HA” and “FT” Option!
A quick review of vocabulary before we go into this subject any further. First when we refer to vSphere, we are talking about the entire VMware ecosystem and all of its components. It is just a short hand for the entire system solution. ESXi is a VMware hypervisor. It is the “host” hardware on which the “guest” virtual machines run on. vCenter is an administrative portal that enables you to manage multiple Datacenters. A Datacenter is a collection of ESXi hosts. I strongly urge all serious engineers to watch Kieth Barker’s presentation in CBTnuggets on this subject, particularly his presentation on HA and FT in the certification training for vSphere VCP5-DCV. Kieth is a truly excellent instructor and he gets paid to make videos!
A more advanced ( read cost more money) strategy for managing server failures in vSphere is either High Availability or Fault Tolerance. Assuming we have three ESXi hosts, lets take a quick look at how each of these strategies would work. Using vCenter we would enable High Availability or HQ at the cluster level. The first ESXi host to boot up, would be nominated Master. Assume the ShoreTel HQ is a virtual server running on this ESXi(1). All the ESXi hosts in the cluster, would exchange heart beats over the management LAN that they all share. Should the heart beat from ESXi(1) not be detected, it would be considered down and the virtual machine would be restarted on the secondary server in that cluster.
An VMware server running VMware Tools, can also generate heart beats between itself and the ESXi host that it is running on. Should the host not receive a hear beat from the guest VMware server, it would consider it down and cause a new instanc of that VM to run. Generally, it is a good practice to use a backup hearbeat to verify the failure of a machine. For example, if the host machine does not generate a heart beat detected by the other hosts in the cluster a back up check could be made to see if the iSCSI storage is being accessed by the missing VMware server. If that heart beat is detected, then the guest is not considered down, but is consider “isolated” and the new instance will not be started.
The issue with High Availability is how long does it take to bring up the replacement guest machine on a new ESXi host? What is the boot time? In Part 2 of this discussion we talked about a configuration that could survive this issue if it was the DVM that went down as the HQ would take over during the down time. If this was implemented in vSphere with HA, the entire process would be transparent to users.
Fault Tolerance is the solution when there can be no down time whatever should the HQ server fail. FT is activated through vCenter as easily as HA, but generates a “mirror” image host that is always running. For example, a ShoreTel HQ server running as primary on ESXi(1) might have a “mirror” host running as a secondary on ESXi(2). Should the primary ShoreTel HQ host fail, within microseconds the ShoreTel HQ mirror or secondary will take over. Not only will it take over as primary, but a new secondary mirror could be started on ESXi(3)!
Clearly FT is the way to go if you feel you can not survive a ShoreTel HQ loss under any condition. Understand that it is resource intensive, as you are minimally running twice the horsepower! Also to keep the “mirror” images alike, you will need a high bandwidth connection between ESXi hosts to provide for “FT Logging” which is all the activity to copy real time between hosts.
As previously mentioned, a copy of VMware vSphere Essentials Kit which includes ESXi for a total of 6 processors or 3 severs with 2 processors each and a copy of vCenter along with updates for 1 year is about $560. vSphere Essentials Enterprise Plus which adds in the functionality of vSphere Hypervisor, vMotion, Cross Switch vMotion, High Availability, Fault Tolerance, Data Protection, vShield Endpoint, vSphere Replication is $4229. Support on all products can be purchase as needed or for term.
Out next project is to figure out how to do this all on Amazon Web Services and at what price!
The video shows key elements of this discussion!