Laptops have always been a challenge to manage, and, frankly, not a lot has changed to address this in recent years. Laptops have all the issues of managing a desktop PC where the user is frequently an administrator on the machine. They are only intermittently connected to the corporate network, making software updates unpredictable, and they are used in un-trusted places, both from a network point of view and the perspective of being physically stolen

Last week Citrix announced that their client hypervisor, XenClient, was available for trial download. There have been ways to run virtual images on client devices before, but they have been based on so-called Type-2 hypervisors where virtual machines run on top of a full copy of a client operating system. These have the disadvantage that there is an additional operating system to manage and to trust. The significance of the Citrix announcement is that it is the first Type-1 hypervisor from a major vendor. Type-1 hypervisors, also known as ‘bare metal hypervisors’, do not require a full operating system, but are instead light-weight isolating layers, requiring a minimum of management and additional security.

Hypervisors allow us to keep the operating system separate from the underlying hardware, and hence allow us to manage each in the most appropriate way. Of course there are components on the client other than the operating system; applications and user environment and the same argument applies to each – they need to be managed in the most appropriate way and virtualization is a convenient way of keeping the components separate. By managing each component separately, it becomes far easier to update each component, and thus ensure that the current standard is always the one in use.

This is a similar set of benefits we currently achieve in hosted virtual desktops running in the data center, and clearly we should expect that the same benefits should apply in client virtualization. But is the client hypervisor the only solution to taming the mobile worker?

In the end this is a two horse race. On the one hand, client hypervisors improve management of the endpoint device. On the other hand, the steady improvement in wireless connectivity in coverage, speed and price make a hosted solution more practical. If we get to a stage where wireless Internet is available everywhere the business needs it, then some of the need for client virtualization goes away.

My phrase ‘everywhere the business needs it’ requires a little more examination. Full global coverage will not be economic for a significant number of years, but most businesses do not need full global coverage, just coverage where they need to do business. If the users you are targeting can get good enough connectivity in all the locations you would want them to work, then that is sufficient. There will be some users where this will not be possible; an example I heard recently was private baking advisors who sometimes have to work with clients on their yachts, outside of coverage now or for the foreseeable future. Clearly there are many less esoteric situations where coverage is either not available, too slow or too expensive.

Which is the right approach? That will depend on individual circumstances but for many job roles this might be the point to review whether the employee really needs a laptop or whether remote access from a home PC would be sufficient. For others, the future might be accessing a hosted virtual desktop and applications over wireless. Many users will still need the ability to run applications locally, and these will be your long term candidates for client virtualization.

It is interesting to note that ultimately the choice of how to serve these users becomes a question of choosing between different types of virtualization: Client virtualization on the laptop itself or presentation virtualization in the hosted example. Ultimately it becomes a question of which are the most appropriate forms of virtualization to use in a particular situation.

Advertisements