Virtualization Feature: High Availability and Fault Tolerance
VMware offers features to ensure that network servers don’t go down, unless you want them to!
People do not like downtime. They like the certainty of knowing that when they want to work, the system is available for their use, remotely and internally. Medium sized businesses of 40 or more users can take advantage of VMware’s virtualization to gain significant benefits in addition to reaching the goal of no unscheduled downtime.
VMware is virtualization software that offers two features that help in this area, high availability and fault tolerance. Virtualization software allows multiple servers to run independently on one physical server. For more information on virtualization and its benefits, please refer to the December 2009 publication at www.triella.com.
Two or more VMware servers working together is called a cluster. For simplicity in this article, let’s call the first physical server Hanoi and the second Singapore. There are virtual servers running on the Hanoi and virtual servers running on the Singapore as illustrated below. Each of the two servers has sufficient capacity to handle all the virtual servers but they have purposely been split between two physical servers so that the firm would not be affected if there was a hardware failure on either Hanoi or Singapore. This is illustrated below.
There are 8 virtual servers in total, 4 on Hanoi and 4 on Singapore. High availability protects against the failure of Hanoi or Singapore. If Hanoi were to fail, all of the virtual servers on it would be started up on Singapore as illustrated below.
Hanoi can then be repaired and then brought back on line. Once on line, four virtual servers can be moved back to it in real time, all without users being aware that anything has happened.
The switchover time for high availability can vary. Detection of the failure usually happens within 30 seconds however it does take time for the virtual servers to start up on Singapore. Virtual servers start faster than physical servers however since there is little time required for hardware related diagnostics during start-up.
If a single virtual server were to fail on one Hanoi, nothing would happen. Manual intervention would be required to start the virtual server up on Singapore or restart it on Hanoi. However, since the server is virtual, the likelihood of failure is slim as it would only occur as a result of a software failure and not a hardware failure.
In the diagram below and as an example, the BlackBerry server has failed on Hanoi. An alert to this effect would be sent to those responsible and the BlackBerry server could then be started up on Singapore or restarted on Hanoi. In this case, there would be some down time unless Fault Tolerance was used.
Fault Tolerance, like High Availability is a feature that allows minimal disruption from a system failure. Fault Tolerance, specific virtual servers are protected. The critical virtual servers are literally mirrored one running on Hanoi and the other in standby mode on Singapore. Thus, when a fault tolerant virtual server fails on Hanoi, it immediately continues running on Singapore. There is no down time. Users are completely unaffected. The Fault Tolerant server is kept in synch with the live server through the VMware software. This is illustrated below:
Fault tolerance is really only for the most critical servers since it does use additional resources. Also, the primary and secondary virtual servers can never be on the same physical server for obvious reasons.
For small and medium sized business that require a network that will require maximum uptime, VMware’s high availability and fault tolerance can help meet that goal.