Which type of virtual desktop will work best for your users?

I decided I wanted to put my thoughts to paper, in a sense, regarding the different types of virtual desktop choices we see in the world today. I will not highlight every technology nuance but, instead, I want to touch on the options at a higher level and talk about some advantages and disadvantages, or tradeoffs. I will not be talking about application virtualization concepts as they are a whole other rat hole and I prefer to keep this post from turning into a novella. That being said, the basic virtual desktop options available today are:

  • Session based
  • Non-persistent
  • Persistent
  • Local
  • DaaS

The first, session based virtual desktop, is based on a long standing technology, originally introduced as Terminal Services. I'll spare the lengthy history lesson for those already aware but know that this technology has been around since 1998 in Windows NT 4.0. This core concept is what powers, what is now known as, Remote Desktop Session Host (RDSH) from Microsoft as well as Citrix's XenApp. Aside from its lengthy history, session based virtual desktops have a lot going for them. One of the main benefits, at least from an IT budget perspective, is the overall costs of session based virtual desktops is lower than most of the other options above. Also, it is very easy to have a solution up and running that supports many users within one instance. Some things to be aware of are application compatibility with a Windows Server operating system and more importantly the compatibility with the multi-session environment of RDSH. Peripheral support is another issue that can arise. It wasn't until later releases of both client and server operating systems that the ability to redirect USB devices to RDSH or XenApp was available.

Non-persistent or pooled virtual desktops are part of the technology stack ushered in by the introduction of VDI in approximately 2006. The main purpose of a non-persistent virtual desktop is to allow users to connect to an instance of a desktop operating system that has been reset to its original pristine state after the last user logged out. The main advantage of this is that users never have to contend with the normal issues that may rise after the desktop is provisioned and continued use. Also, the challenge of application compatibility with the server operating system and multi-session environment is mitigated. But non-persistent virtual desktops produce their own challenges. One being an ability to persist user initiated changes, such as application configurations, requires additional technologies to manage the users' profiles. Other changes, such as user installed applications, usually require yet another technology layer to ensure their persistence as well. One thing to keep in mind is the cost to deploy non-persistent virtual desktops is usually greater than those of session based desktops.

VDI also includes persistent virtual desktops in the fold of options. These virtual desktops do not carry the issue of needing to insert other technology layers to persist the user settings and applications. Users are able to make their changes and, as the name suggests, said changes will persist. Use of traditional desktop management solutions is also more easily facilitated by this type of virtual desktop technology, making them align more tightly with the existing physical desktop management strategies. Where non-persistent desktops removed the traditional management in favor of an always pristine image, persistent desktops trade that in for flexibility of management and a certain amount familiarity. Costs for persistent virtual desktops are a factor though. Due to the expanded storage requirement, the price of this model is greater than that of non-persistent virtual desktops. There are storage technologies, like deduplication, that can reduce this expense but the increase in total cost of ownership will still exists.

Local virtual desktops, also known as offline or client side virtual desktops, were introduced a few years later and provide yet another avenue for the implementation of desktop virtualization. Two options for this are type 1 and type 2 hypervisors and each has its own unique benefits and drawbacks. Where the previous desktop virtualization technologies require constant connectivity for virtual desktop access, local instances are not bound by the same standard. Since they are indeed local to the client, the virtual machine is always available. This proximity to the virtual desktop removes any performance issues associated with connectivity across the network. A disadvantage to this approach is that the client machine requires the native horsepower to run the virtual machine. The hypervisor and virtual machine consume and share the same physical resources so naturally this could reduce the overall availability of required resources for the users' operating system and applications.

Desktop as a Service or DaaS is the final desktop virtualization technology on the list. In reality this is really a culmination of the first three technologies with added consumer style features. Service providers can offer session based, non-persistent, or persistent virtual desktops to the end user, on demand. DaaS decouples IT from managing the underlying infrastructure required to deploy virtual desktops and allows for an organization to look at desktop virtualization as a completely operational expense. It's usually as simple as going to a web interface, requesting a new virtual desktop, and the rest is done behind the scenes. Being that this is typically a hosted solution, the users' data is not always in close proximity to the virtual desktop itself. This can pose performance problems depending on how often data is read and requested by the end user. There also can be licensing restrictions which can make the overall cost of consuming DaaS expensive if the base requirement is to only consume a small number of desktops.

The thing that I love most about this topic is what is typically heard from vendors. Their message about which technology is best for a customer or their environment will be the one that plays to their strength as a solution, not necessarily the one that meets the business or technical requirements. Once you weed through the nonsense and FUD, you can find what truly works for your users.

Comments