IT pros like to have well-defined outcomes from new technologies, but business managers like to know that IT isn't needlessly reinventing the wheel. That dynamic tends to drive us toward reference architectures as a recipe for VDI deployment.
The problem with a virtual desktop infrastructure (VDI) reference architecture is that the recipe may never have been tested before. Plus, you need to determine whether you actually want the dish that the recipe produces.
More reference architecture resources
VMware View reference architecture overview
Reference architecture scaling for Citrix
Building a private cloud with a reference architecture
Understanding the use case and scale that a particular reference architecture (RA) is designed for is crucial to getting the outcome you want. You need a VDI reference architecture that has been field-tested, because lab-built ones are usually only fit for lab use. Often, it's best to use an RA simply as a learning tool to help you design a better VDI project.
Here are three important questions to ask yourself and your team before using a VDI reference architecture:
What is the reference architecture for?
Most administrators use an RA to solve a particular problem. For instance, a group of mobile sales professionals requires 3G access using laptops in the field, or a nurse working on a ward with thin clients needs access over a secure network. To determine if a VDI reference architecture is necessary for your project, you need to identify the use case you are trying to accommodate and see if the RA is suited to that case.
How big or small does the RA go?
A reference architecture usually has a target scale or at least a minimum scaling unit. Good ones often work with a building block for scaling, so if the RA scale unit suits your VDI project, you may be able to use it. If you need 200 desktops and the architecture scales in modules of 2,000 desktops, however, then it will be hard to make your requirements fit. Similarly, if the scale unit is 200 desktops and you need to support 7,000, that won't work either.
Where did the VDI reference architecture come from?
A good VDI reference architecture comes from a real deployment with a real customer: Someone has built it and ironed out the unforeseen issues. Hopefully it has been in operation for a while and the support team has learned from user feedback.
However, some RAs are purely a thought exercise: How would I go about building a VDI project for a particular use case? These architectures have never been built and may not actually work as they are envisioned. Others have been built in a lab and validated with standardized, automated workloads rather than real users. These RAs often aren't easy to operate and support, and they never learn from real use. For instance, things like printing may never be needed or tested in a lab.
With these three considerations in mind, you should also remember that a VDI reference architecture doesn't have to be an exact recipe. The chance of finding a field-tested RA that suits your use case and scale is fairly small. A better approach is to read a few RAs and use them to learn how to build different environments.
Using the knowledge gained by RA designers helps you get a deeper understanding of the VDI product you plan to use. VDI reference architectures tend to be written by smart people with access to quality technical resources -- often the VDI product vendor or actual infrastructure components.
This was first published in November 2012