The Re-emergence of Citrix XenApp and Microsoft Terminal Services

Brian Madden just blogged about the re-emergence of Terminal Servers and Server-Based Computing (SBC) and I tend to agree with his views. See Brian’s article titled “The inverse bell curve of Terminal Server / SBC: This stuff is going to be huge again!” here.
The term Virtual Desktop Infrastructure (VDI) has been widely used to refer to Hosted Virtual Desktops – i.e., desktop VMs on a virtualization platform, with the desktops being assigned to users as needed. Over the last couple of years, VDI have attracted a lot of attention – almost to the point where solution architects feel that VDI is the answer for any remote access requirement.
Real-world experience with VDI has been different though:
    • VDI has proven to be more expensive compared to more conventional remote access mechanisms such as XenApp and Terminal Services. With VDI, each user has his/her own desktop, their own operating system and applications running in the OS. Not only does this increase the resources (CPU, memory, etc.) required per user, it also has a higher licensing cost because of the need for one operating system per user.
    • While many pilots and early deployments have proven that VDI works, achieving the degree of scaling that is necessary to make VDI commercially attractive has been a huge challenge and an inhibitor for wide-spread adoption of the technology. As enterprises try to squeeze more users per physical server, they have to deal with unhappy users who are experiencing performance problems and slowdowns. A lot of time is spent in figuring out where the bottleneck is. Often, knowing where the bottleneck is may not be sufficient. This analysis may reveal that the initial sizing of the infrastructure was not done right and a far more significant investment may be required on the storage or the virtualization front, which in turn adversely affects the financial viability of the project.

In contrast, XenApp and Terminal services are very mature technologies that are only getting better with time. With a smaller per-user footprint, these technologies have proven to deliver a far greater number of users per physical server than what the VDI deployments have delivered

  • VDI has turned out to be harder to manage than XenApp and Terminal Services. Firstly, there are more infrastructure tiers and more software and hardware components involved (the connection broker, virtualization platform, storage devices, web front-ends, individual virtual desktops, software agents for the connection broker on each desktop, etc.), and these technologies are far less mature than the equivalent technologies for XenApp and Microsoft Terminal Services.  Early VDI adopters have been finding issues with these technologies as they try to stretch the usage of these technologies. Further, for managing VDI, most enterprises have tried to use the same tools they have been using for virtual server management. Refer to this discussion on why a virtual server management solution cannot be directly applied to manage VDI.

Clearly, the growth of VDI has not quite matched what industry analysts had predicted earlier this year. One of the major reasons for this is while VDI is an ideal solution for environments where separate operating systems are required for each user (e.g., because different users are accessing legacy applications that cannot work together in the same operating system, or because the users require access to modify the operating system for their specific needs), VDI is often an overkill if all that is required is remote access to applications hosted on a remote server. XenApp and Terminal Services are proven technologies for such applications.

Many enterprises are learning this mid-way through their efforts at moving to desktop virtualization!