Virtual Desktops: Building a better mousetrap

When it comes to building a mouse trap, what is the first thing you think about?  Chances are, you don’t think about the gauge of wire you’re going to be using, or the tension rates for the springs.  If you’re like me you think about the base needs of the mouse, and how you’re going to address them in order to attract the mouse to it’s new “home”.   What does the mouse need?  Food, water, shelter, etc?  When it comes to designing a virtual desktop solution we too often see the opposite.  People start designing a solution by identifying the server hardware and storage infrastructure and then designing from there.  Cores and IOPS, memory and network throughput.  But we’re forgetting about the base needs of the beings for which this environment is being designed and built; the end users.  In order to attract end-users to the solution, and ensure that they will fit within it, you need to start with the end user’s requirements and work down until you reach the hardware requirements.

At Citrix Synergy 2013 in Anaheim California, Niraj Patel and I delivered a session on best practices and methodologies to consider when designing a virtual desktop solution.  Here are some of the design considerations from that presentation:

When thinking of virtual desktops, consider the environment in five layers:

• User Layer: Endpoints and required end user functionality such as multimedia
• Access Layer:  Connection between end users and their virtual desktops
• Desktop Layer:  Virtual desktops, applications, profile, personalization and policy
• Control Layer:  Infrastructure controllers required to support the environment
• Hardware Layer:  Physical devices; servers, storage and network required

As a design principle, start with the User Layer and work your way down.  Identify the users needs in terms of application and functionality requirements, what devices they will use, and how they will access the systems.  Group the users based on this information, and group the applications as well, so you can map applications to users.  User and group mappings should be established during the Assessment phase so that appropriate design decisions can be made during the Design phase.

Once user-level design decisions have been established, move down to the Access layer, and identify how users’ access needs will be addressed in the design.  Are user groups accessing the environment locally or remotely (or both)?  Will they roam and use multiple devices such as tablets and smartphones in the enterprise environment?  What access policies should  be put in place?  These are just some of the considerations.  Once the design decisions on the Access layer are determined, we can establish the access layer elements such as StoreFront and NetScaler Access Gateway that are required.

Similarly, continue through Desktop layer, identifying the decisions around personalizaton, profiles and desktops, and down through the Control layer, identifying the needs for scalability and control element design.  Once you have all that together, you can put together the requirements for the hardware, considering the overall design in terms of modules and scalability required.

The top down approach allows you to design the solution from the same mind set as users who are accessing it; from the end-points to the virtual desktops or applications running on the hardware in the data center.  Doing this will help to make sure that user requirements are foremost in mind during the design process.

Source : Citrix's blog