In my network consulting practice I am often working with clients who are deploying video conferencing. One of the first steps in this process is to determine the video demand, which means predicting how much bandwidth the video conferencing will use on the Enterprise network. One of the critical components of this calculation is to understand how often users will be using the video conferencing equipment.To get a better handle on this question, I had a discussion with Jim Idelson of DesigNET. Jim's consulting practice focuses on the application and process side of video conferencing, looking at usage patterns, service quality, operational efficiency, and the business impact of video conferencing deployment.
Jim pointed out that we have to look at the room-based video conferencing and the personal video conferencing as two quite different models, from a utilization and efficiency point of view. The room-based model, whether it be traditional video conferencing, high-def or Telepresence, is all based on having a shared resource (typically a conference room equipped with a fair amount of expensive gear) that is scheduled. Conversely, personal video conferencing is just that--personal--meaning that the equipment is dedicated to one person (either a low-cost desktop appliance or software and a web-cam on a PC). The utilization of room-based systems is quite important and heavily dependent on the efficiency of the scheduling process, and on the size of the user community that those systems are intended to serve. The utilization of personal systems is only dependent on the frequency with which the individual user wishes to use video or has an opportunity to use video.
Jim and I are both interested in video utilization, but for two different reasons. And it turns out we need to look at different numbers as well.
Jim's interest is in understanding both overall and peak utilization of resources, in order to manage the quality of the end-to-end user experience, while keeping the cost of providing the service as low as possible. So he is calculating the investment and operating costs distributed over the number of video-conferencing hours.
I am trying to understand the demand on the network, so I really need to understand the concurrent utilization during the "busy hours" for specific topological portions of the network. As an example, I need to know how many of the systems in the Chicago office will be used concurrently, and if they will all be using bandwidth on the Chicago MPLS access link. If calls are made within the Chicago office (or complex, or campus) then there may be plenty of bandwidth, whereas if most of those calls are to remote offices, the WAN access link may become a significant constraint, as it is the more scarce resource.
Jim and I worked up a couple of models for how to think about video conferencing utilization for these two classes of video (room-based and personal). I'll discuss these in the next couple of blogs.
