My desktop is always there for me
High availability loveliness
Desktop virtualisation can be an asset to an organisation, but what happens when something goes wrong? In a conventional non-virtualised desktop system, users with a downed network or crippled server could perhaps still work on something locally.
But when that same downtime means that your whole desktop vanishes, high availability becomes even more important. How can we keep the virtual desktop up and running?
For the naysayers, it's worth pointing out that, done well, desktop virtualisation provides a higher level of disaster recovery than local desktops. Even in mid- to large-sized enterprises where IT is more locked down, many PCs may not be backed up.
Alternatively, with a VDI system, at least the system is available centrally if the desktop device gets fried. "You can connect via a network to the last session that you had, and it's easy to recover. All you're doing is connecting to the session, with the data and the applications," points out Mike Osborne, managing director of ICM, a business continuity provider, adding that this can get individuals up and running faster than before in the event of a desktop crash.
The other upside to VDI is that it enables home working. "What you're talking about is the ability to enable 100 per cent of an organisation's resources in a disaster instantaneously," he says. That assumes, of course, that you can get corporate telephony out to the remote desk, too.
Erwin Vollering, service director of virtualization at Glasshouse Technologies, a cloud services provider, starts by asking what the client's uptime requirements are. You'll often find that different departments have different requirements, he explains.
"For some, it needs to be available 24 x 7, no matter what," he says. "Then we look at a solution where we can facilitate that, in which certain brokers can do that for user groups that have a higher need than others."
When this is ascertained, the implementation team needs to assess every link in the chain between the desktop client and the back-end virtual machine. This includes the network connection from the desktop to the data centre, or wherever the virtual machines are being hosted. It includes home users' gateways, routers, firewalls, and SSL VPN concentrators. Even the hypervisor has to be designed for extra capacity for failing over from another box.
Synchronising a storage area network is difficult enough, but even synchronising locally stored data on the server between different hypervisors can be tricky. Part of the problem is taking a variety of different desktop images, and synchronising them along with the user data that also has to be available.
"If you're using stateless desktops with a profile virtualisation solution, the amount of data you need to keep in sync between sites can be reduced," says Russell Raynsford, managing director of infrastructure services at Molten Technologies.
Stateless desktops are non-persistent, meaning that they get destroyed every time the user logs off, and reinstantiated for the next session. IT administrators can give their users the best of both worlds by using products to store a user customisation layer, containing all of the changes that the user has made to the desktop.
This means that a shared desktop image can be stored for the user, separately to the far thinner customisation data layer. The result: far less data to replicate, leading to a reduction in SAN or local storage costs.
The alternative for the strong of heart is simply to offload the whole sorry mess onto someone else. "If you have an external virtual hosted desktop provider, you only need the almost universally accessible internet to get to your desktop," says Raynsford. "This can be a plus point for building disasters, such as an office flood, or if nobody can get there because of snow." External providers can also pool disaster recovery and share the costs across customers, he points out.
The same is true of application virtualisation systems such as Microsoft's App-V. It streams code to the local client, which then enables users to continue operating in the event of a network outage.
The danger with third party hosts, of course, is that they may not provide IT administrators with the level of configurability that they require. You'll need to assess how flexible the administrative interface is.
IT teams must also look into pricing mechanisms to see how the costs of software licenses vs rented operating systems and applications would stack up . It’s also worth keeping in mind what the licenses allow you to do. In the world of Windows for instance there are special licensing requirements for accessing Windows running on a server.
Don't underestimate the cost of making a virtualised desktop infrastructure truly resilient, then, but don't underestimate the things that you can do to help mitigate the tasks involved. ®
Sponsored: Hyper-scale data management