Much of the emphasis on disaster recovery planning is understandably on servers and server workloads. Organizations...
that have adopted VDI must consider the role that virtual desktops will play in disaster recovery, however.
When disaster strikes, it is important to restore everything to working order as quickly as possible. Users need to do their jobs, and customers must be able to make purchases, contact support or get help with problems.
Prior to desktop virtualization, endpoint devices often received minimal consideration with regard to disaster recovery (DR) planning. IT shops often assumed that a vendor would be able to supply brand-new desktops on an as-needed basis. Because desktops do not always contain corporate data, administrators understandably focused their attention on back-end resources.
Virtual desktop infrastructure changes everything. If an organization has fully adopted VDI, users simply cannot access the tools they need to do their jobs outside of a virtual desktop. Thus, businesses must have a plan in place for quickly getting virtual desktops up and running after a major disaster.
There are a number of ways to ensure that virtual desktops remain readily available after a disaster. There are advantages and disadvantages to each method, and some are easier or less expensive to implement than others. None of these methods is an ideal fit for every type of organization, so it makes sense to examine several different ways to do VDI disaster recovery.
Hyper-V on the desktop
The first method isn't commonly used, but I do know of at least one organization that has chosen to use it. The organization in question currently hosts its virtual desktops on Microsoft Hyper-V. It also stores a backup copy of its virtual desktop securely in the cloud.
In the event of a large-scale disaster, the organization has an agreement with its hardware vendor that says the vendor will lease a number of desktop PCs that the company can use until it fully recovers from the incident. According to the agreement, these PCs will run Windows 8 and have Hyper-V installed. The organization's plan is to push a copy of the virtual desktop to each PC and use Hyper-V to make the virtual desktop available to users.
The amount of work required to put this plan into motion is probably impractical for larger companies, but it could prove to be an effective option for smaller organizations. This method allows a company to get virtual desktops up and running without any dependency on back-end infrastructure.
The only real requirement is a Dynamic Host Configuration Protocol server that can allocate IP addresses to the virtual desktops. For that, the previously mentioned organization is relying on a wireless router that provides network connectivity to the PCs and that supplies IP addresses.
Windows To Go
One organization that adopted this plan created a large number of USB flash drives in advance of a disaster. Those flash drives are stored off-site in a secure facility and will be distributed to users in the event of a disaster. Users will be able to boot their own personal laptops from the USB flash drives and access a fully provisioned corporate desktop.
Unfortunately, using Windows To Go isn't an option for organizations running Windows 7, but companies can use Boot to VHD as an alternative.
In either case, the USB flash drive's capacity will limit the size of the virtual desktop image. Desktops that are provisioned with a large number of applications may not be good candidates for use on a USB flash drive.
The other disadvantage to this method is that to be truly effective, the flash drives have to be prepared ahead of time. This might not be a big deal if the virtual desktop images remain fairly static, but it could quickly become impractical for organizations that refresh their virtual desktop images on a regular basis.
One of the more common methods for VDI disaster recovery involves building a deployment that spans multiple data centers or that stretches to the cloud. The actual logistics to do so vary considerably depending on vendor. This tends to be not only the most reliable, but also the most expensive option.
The basic concept involved in spanning data centers is that the host cluster where the virtual desktops reside is stretched so that it spans multiple data centers. The storage device used to store the virtual hard disks is replicated to the secondary location. That way, copies of the virtual desktops exist in both places.
Although it is theoretically possible for virtual desktops to fail over to the secondary data center, it may be more effective to create an entirely separate pool of virtual desktops within the secondary data center; hosting virtual desktops in an alternate location changes the networking requirements.
In some cases, it might be easier to connect users to a virtual desktop that was specifically provisioned for use in the alternate location rather than trying to reconfigure existing virtual desktops to work in an offsite location.
Offline virtual desktops
VMware offers a feature that allows virtual desktops to be checked out by mobile users and used offline. Companies could theoretically use this capability for disaster preparedness in the event of advanced notice of an impending disaster, such as an approaching hurricane.
The disadvantages to using this technique are fairly obvious. First, it probably isn't something that's going to be easy to implement after a disaster has already occurred. Second, this feature only works in VMware environments.
Organizations that have adopted VDI must address virtual desktops within their business continuity planning. Keeping back-end server resources online after a disaster offers little benefit if users do not have virtual desktops from which they can connect to those resources.
How to recover virtual desktops
Using the cloud for VDI DR
Guide to VDI DR