Companies moving from traditional PCs to virtual desktops need to plan for a significant increase in VDI network traffic as applications and user data traverse their LAN and WAN connections to remote end users.
To prepare your network infrastructure for a large number of virtual desktops, you need to choose the proper display protocol, add the right amount of bandwidth and set
Virtual desktop infrastructure (VDI) depends on end-to-end stateful communications between the client device -- for example PC, tablet or smartphone -- and the back-end infrastructure. The term stateful means that the client is always communicating back to the virtual desktop environment. In essence, the communication is very much like applications such as Voice over Internet Protocol (VoIP).
An unreliable network can result in dropped packets, which can significantly degrade the user experience. In the case of VoIP, telephone calls would sound garbled and lose their connection. With virtual desktops, it can result in poor screen updates and unresponsive mouse clicks and keyboard activity.
More on the VDI network
How to manage VDI network bandwidth
How VDI can make the network more secure
VDI strains data center networks
A good VDI network should offer a consistent stream of data. When admins look at their networks, they often focus too much on raw throughput. What really matters is a stable and reliable network that doesn't drop packets. A sound network should already have the fault tolerance necessary for VDI implementation -- including teamed network interface cards (NICs), multiple switches, and more than one backbone to the server room or data center.
These components are probably the strongest links in your infrastructure, but don't take them for granted. You'll want to ensure that your infrastructure will continue to function after an NIC, physical switch or server-room failure. Few organizations, however, conduct regular hard tests of their physical infrastructures.
In addition, it's important to have the correct monitoring tools to identify whether N+1 resilience is functioning. Some network resilience is so seamless that when Component A fails, users do not notice -- until Component B fails as well. Conduct an audit of your monitoring tools to confirm that you'll be alerted in the event of a component failure.
Once organizations have ensured that physical redundancy and monitoring are functioning properly, they should focus on the weakest link in their environments -- which is typically the wide area network (WAN). This is especially important if your organization aims to offer virtual desktops to end users at home as well as in the office over the VDI network.
If you think the majority of your users are Dilbert-like folks who sit in a cube at the same computer all day, think again. User and management expectations are changing. Even users who are not "road warriors" will want to catch up on work over the weekend or work from home on some days. They don't expect to always be in the office when they're working.
At the very least, you should consider two separate pipes to the Internet at the site where your VDI environment is located. These should be with two distinct service providers; ideally, their infrastructures should be separate as well.
In the world of the WAN, it's not unusual for providers to resell one another's infrastructure, which could allow two apparently separate communications providers to both be affected by an outage further upstream from your infrastructure. Sit down with an Internet service provider before signing a contract to negotiate a proper service-level agreement (SLA) that guarantees a defined quality of experience. The SLA should include penalties if the provider fails to deliver the service to the agreed-upon standard.
This was first published in October 2012