Desktop virtualization is clearly a hot topic in IT, but a closer look reveals that some elements are still missing and that in many use cases problems would be better addressed using “classic” technologies such as Client Lifecycle Management and terminal services.
That certainly doesn’t mean that there is no need for desktop virtualization. There are several interesting use cases for desktop virtualization and the technology will become more important over time. A successful deployment of desktop virtualization, however, requires a well-thought concept and careful planning, and currently only fits some use cases.
Moreover, desktop virtualization has to be understood as a complementary concept that in most situations will exist alongside other desktop concepts such as terminal services or “fat” desktop deployments managed with Client Lifecycle Management tools.
A complex backend, a lot of storage
When looking at desktop virtualization, it’s first essential to understand that every approach builds on quite a complex backend infrastructure for the virtualization itself and its management, and requires sufficient storage.
There is no simple deployment of desktop virtualization. The approaches of any vendor—Citrix, Sun, and VMware the leading ones; others including Parallels, Quest, or Microsoft add solutions or are on their way to entering the market—require a lot of planning and deployment for the entire infrastructure.
Compared to the other approaches for managing desktops, including terminal server environments or Client Lifecycle Management tools, implementing desktop virtualization is considerably more complex.
Despite the efforts of vendors to reduce them by reusing parts of the virtual environments and regardless of the chosen approach, desktop virtualization requires massive amounts of storage. The bandwidth requirements for a fast deployment of virtual machines to client systems also shouldn’t be taken lightly in case that mobile and disconnected use cases shall be supported in the future. And if protocols like RDP and ICA are used for remote access to the server-hosted virtual clients, there are similar restrictions (like in terminal server environments).
Don’t underestimate management efforts
The same is true for the management of desktop virtualization. All vendors provide some tools for managing the virtualized environment itself and virtual machines. Providing that there are a very limited number of virtual machines, these tools will be sufficient.
Once the customer considers using more varied configurations of virtual desktops, however, management becomes a real issue. Flexible mechanisms to manage the different images are frequently missing, and most vendors don’t integrate well even with the existing tools for Client Lifecycle Management. Although there is good support for customizing terminal server environments today, there are only a few solutions for the desktop virtualization market. Even the management tools aren’t (at least in most cases) fully mature, with several tools usually required for managing the environment.
Desktop virtualization is a relatively new market segment, thus this immaturity is no surprise. Given the complexity of the problem, though, more advanced and better integrated tools are mandatory.
And what about mobile users?
The support for mobile and offline users, e.g. users who aren’t always connected , is another open question. Most vendors currently are in the beta phase of developing and deploying solutions for those types of users. But it is obvious that it will take more time to finalize these solutions – and as said above: The bandwidth requirements shouldn’t be underestimated.
Trends
This broader support for different use cases, including mobile users, is one area where we observe significant improvements. Another is the support of different operating systems that can be deployed in virtualized environments.
There are also more vendors trying to enter that market. But since there are very few companies with their own strong virtualization technologies in the market, the number of core vendors will remain limited. There will be additional offerings built on top of solutions from VMware and other vendors, but given the complexity of virtualization and the very small number of developers of core virtualization functionalities worldwide, market entry is difficult.
In terms of the present shortcomings of desktop virtualization, other areas where vendors are trying to improve their current solution are in storage requirements, in mobile support, the support for video and printing,, and in the ability to support individual changes per user in these virtual machines.
The next major trend will be better integration with technologies for application virtualization. In a virtualized environment, this appears to be the logical approach for deploying applications to some of the virtual machines. Most vendors of desktop virtualization also provide technologies for application virtualization. There are, however, no really integrated solutions currently available in which any aspect of virtualization can be managed from one interface. We expect that to change within the next 12 to 24 months.
Within that timeframe we also expect to observe more solutions that provide virtual machines over the internet, e.g. from the “cloud.” That will become a more important market in the midterm, when, for example, desktops for small businesses are distributed pre-configured by specialized cloud providers.
Overall, desktop virtualization (and, even more so, application virtualization) as cloud services is amongst the most promising new use case in IT. Higher bandwidths and an optimized support for offline use cases with some synchronization of changes will be required to support these use cases. These will become reality in the near future.
The strategic relevance of desktop virtualization
The existing shortcomings of desktop virtualization raise some questions: Will desktop virtualization ever make sense? When and how should an organization deploy desktop virtualization today? What strategic impact will desktop virtualization have over time?
To start with the first question: Yes, desktop virtualization is a valuable technology—but it shouldn’t be overhyped. Desktop virtualization is, even today, an interesting technology. Today it is limited to internal networks with always connected and standardized clients. Some of these limitations will be removed.
Nevertheless, even in the future organizations will need to analyze whether terminal service approaches are still a better solution for customers. They are proven, and there are advantages for both types of technologies. We expect that, especially in large environments, both approaches will frequently coexist.
Besides this, there will typically still be “classic” desktops, which are deployed locally and managed by Client Lifecycle Management tools. These may be replaced by better management of virtual desktops over time.
Form a strategic perspective, desktop virtualization is an important technology. But it has to be understood as one of several approaches for managing desktops as well as the multitude of use cases, from kiosk users to users with standardized desktops, and power users or developers with very specific requirements, which usually requires different approaches to desktop management. The relationship between application virtualization and desktop virtualization also has to be understood.
Desktop virtualization will gain momentum when these desktops are delivered from the cloud by specialized service providers. Thus a strategy for desktop virtualization has to be aligned with a cloud strategy.
Given the complexity of desktop virtualization, it cannot be implemented tactically and always requires a well-thought-out strategy for overall client deployment and management. Deploying desktop virtualization tactically is extremely expensive. Organizations should therefore consider desktop virtualization as an element of their IT, and should first define a strategy and build a roadmap for future client deployment and management before implementing desktop virtualization solutions.