VDI: “Killer App” for the Network, or Network Killer?
A lot of my coverage over the past couple of years has been focused on the increasingly interesting mix of applications that get delivered over enterprise networks and the impact that has on how networks are being planned and operated. This cocktail has become increasingly volatile, as latency-insensitive traffic loads are mixed with a growing number of latency-sensitive traffic loads. Chief among the latter has been VoIP, but the growth in live, interactive videoconferencing cannot be discounted. In my research from 2011, Videoconferencing Impact on Network Management, I found that network practitioners found a number of issues at the root of performance problems with videoconferencing, chief among them latency and packet loss both in the WAN and, surprisingly, the LAN. Most organizations are dealing with this by applying network optimization and QoS policies.
Enter upon the stage another latency-sensitive application: VDI. Virtual Desktop Infrastructure is all the rage as a means to reduce security risk around data leakage in the face of inevitable BYOD and endpoint consumerization trends. There are a number of other driving business reasons for embracing VDI, but BYOD is, from my viewpoint, the most compelling potential tipping point.
Now VDI is not everywhere yet, but interest is high. During my recent Network Management Megatrends 2012 research project VDI turned up as the second most-influential broad IT initiative in the minds of IT execs when it comes to networking and network management. The only higher response was for server virtualization. Curiously, networking pros were much less concerned – half as much as the execs – which indicates that adoptions are still early and the full effects have perhaps yet to be fully appreciated.
But why should we care about VDI from a network perspective? There are multiple reasons. First and foremost, separating the human interface, where mammal meets plastic, from the actual instance of a desktop application execution environment, which is now hosted on a remote server, introduces a new performance variable – the network. Secondly, VDI performance relies on network architecture and plans in the back office /datacenter as well. And lastly, hosted desktop services via VDI technologies (a.k.a. Desktop as a Service, or DaaS) is a growing field, and can never be fully assured without some degree of mindfulness about the role and influence of network access.
In terms of delivering an optimal VDI experience between the VDI server and the client, the challenge primarily revolves around network optimization and traffic prioritization. And very important in this is consideration of exactly where the client sits. Many early VDI deployments have achieved success in part because they are restricted to LAN/campus environments. But the greater challenge is to extend the benefits of VDI to clients that reside remotely, either across a corporate WAN or via broader Internet access. Here, traditional technology solutions such as application delivery controllers (ADCs) and WAN optimization controllers (WOCs) can be applied and can add significant value. Take a peek at the noise that F5, Array, and Silver Peak are making around this. Another valuable approach is to look at Internet optimization services to improve delivery effectiveness – see Akamai’s Terra Solution if you are looking for examples here.
On the backend, in the data center, there are also plenty of opportunities to improve and optimize the networking aspects of a VDI deployment. In particular, VDI servers host the desktop sessions, but are essentially front-ends for traditional multitier applications. Further, the access adjacency of storage is hugely influential in the total VDI performance picture. Done properly, the end-user experience can be even better than a physical, native desktop. Done improperly, that same end-user will be substantially degraded. For more on this, take a look at what the VCE Coalition and V3 Systems are doing.
Later this month, I am launching a primary research project into the best practices for planning, monitoring, and optimizing networks to support VDI deployments. In particular, I’ll be looking for clear emerging trends and requirements so we can all learn from those who go before us. If all goes as planned, I will have results to share by the end of June. So stay tuned! The results could truly be “killer”…