Desktop Virtualization in 2009

Applying VMware allows the operations organization to eliminate a lot of conversation time by skipping over painful integration efforts including normalizing on a common operating system release and service level, re-certifying the application software, looking for ways to normalize diverse service level agreements and adding new systems while the old ones had excess capacity. Many IT executives are getting a lot of back pressure from their CFOs when it comes to buying more systems - when most of the ones installed aren't even close to be fully employed. 

 

Using VMware, especially ESX, enables rapid portability of the applications and their environment, intact, to systems that support VMware's ESX enabling the operations organization to move the applications around, quickly test them out, and move the applications around. It separates the systems hardware, the operating system and application, and enables all sorts of "why can't the 'application' run in an internal or external cloud," assuming the economics and service levels, security, and data controls play out.  The appeal to this has already been tested with Software as a Service (SaaS) applications such as pioneered by Saleforce.com.

 

Since virtualizing application systems (the application and its environment) has been so helpful, many companies are looking at how to apply the same approach to desktops to achieve similar cost savings and flexibility using virtualization solutions from VMware, Microsoft, Citrix and others as well as move the desktop into the cloud. 

 

Making this leap has not proved quite as straightforward. To date, as Charles Babcock's article outlines, there are numerous and diverse approaches being developed and implemented. Only the pragmatic have had relatively satisfactory results. The primary reason is that the problem being addressed is not the same. The application is a one-for-many solution that can be implemented transparently to users without major infrastructure changes while most virtualizing desktop solutions are very visible many-to-many issues. What I mean is that the number of components being managed by virtualizing a data center goes down, if only by the number of physical systems and the applications are not touched. The economics are clear. Bringing desktops into the data center is a different proposition since it involves many infrastructure and end-user changes, and careful attention to the nuances is required.

 

For example, as an over simplified comparison the original desktop environment includes: a PC and all its software, client components associated with server side applications and VPN software to access the applications. The initial thinking is that moving the physical desktop environment with PC environment with remote access portal to data center-side software or a thin client portal, a connection broker, and a virtual desktop image and image management and the user needs to be retrained. This doesn't look too exciting in a down year or doubtfully in any year. 

 

The subtle differences between thin clients and PC edge devices gets ensnarled with connection availability requirements, the degree of savings and training associated with end users and the complexity associated with managing the software/applications on their edge device. A reasonable question that is being asked by some IT executives is: "if this is the answer, what was the problem?" I believe the problems that need attention in a down year have to do with reducing CAPEX and OPEX without an initial investment bubble. This can best be served with seamless transparent simplification and without added complexity and major investments and is why cloud computing pay as you go has appeal.

 

Putting on my old IT operational hat, I would go back to a terminal concept e.g., a host with users logging into their Virtual Machines running on a shared server. This could be done seamlessly and lets one break the technical challenges into more manageable components: a "terminal", a network that gets the authenticated and securely connected user to their application (e.g., virtual desktop) and support services. 

 

Having sliced the problem into three manageable pieces: user support, network support and data center-based desktops, it is easier to see the economics that have a relatively clear understanding in order to align with the business goals, for example in 2009, this approach enables the IT organization to start the "desktop virtualization" effort using existing tools while they experiment with various efficiency mechanisms to squeeze more efficiency out of the desktop content management and resource consumption.  

 

If it were me, I would not spend a lot of energy on introducing special support for mixed environments or deploying interim thin client hardware unless I can see it applicable for three to five years. 

 

For laptop and desktop users, I'd give them a choice so they can repurpose their current edge equipment into a terminal or they can run in two modes (PC and terminal mode). 

 

For new mobile users or those who work from home, I'd offer them the ability to use their personal laptop or desktop with the terminal portal or provide them with a zero touch terminal device. Zero touch means that when you power it on, it connects to the network. Once the user logs in from their terminal, they have access to their virtual desktop if it were local.  

 

I think that this approach lines up with the comment at the end of the article regarding that the users (employees) did not really want a laptop but rather the remote access.

 

If you are interested in benefiting from cloud computing and virtualization and the above scenario makes sense, you will want to look at the SIMtone approach. SIMtone enables you to get comfortable with desktop management considerations while actually seamlessly delivering the service in days without the investment bubble.