The need for supreme availability for VDI users is subjective. Your strategy will depend upon the criticality of the work and the potential financial loss from any disruption.
Most organizations achieve high availability through implementing technologies like failover clustering and load balancing. Should hardware fail, users are still connected as the workload switches to another available host.
There are several important considerations around VDI and high availability. Perhaps the primary question to answer is, what happens if a user loses their VDI session in the middle of a task because the server hosting the VDI session fails and the task needs to be restarted on a second VDI server?
While no one likes to lose access to their work, there’s a big difference between a set of call center employees using VDI to access customer service software and engineers using VDI to architect complex pieces of equipment in a CAD program or a financial adviser needing real-time access to market data to serve clients and make quick decisions.
In the first case, loss of the VDI session is inconvenient, but it’s also unlikely to cost the organization a substantial amount of money. Each call center task rarely takes more than a couple of minutes to resolve. If there’s a disruption, restarting it with a new VDI session will cause a few more minutes delay as the call center workers reconnect to new sessions and restart the process of resolving the customer query.
In the second case, involving more complex work and expensive employees, financial losses can be higher. It might be that each engineer has a personal virtual desktop which remains unavailable until the server is repaired. It may be that all changes since the last save are lost forever. For these types of users, it makes sense to configure the VDI deployment so that the sessions are always highly available, such as by setting up a failover cluster to host VDI virtual machines. In the event that the VDI server fails, the virtual machines automatically shift to another functioning VDI server node in the cluster, preventing any noticeable work disruption to the user.
Determining the potential ROI of implementing highly-available VDI systems is a sound first step. If the loss of VDI virtual machines due to server disruption would cost the organization thousands or tens of thousands of dollars, then deploying a highly available VDI architecture is a wise strategy.
Underwritten by HPE, NVIDIA and VMware.