At Interop 2009 in Las Vegas, track chair John Bartlett of NetForecast Inc. will be leading a series of video conferencing sessions covering network support for video conferencing, HD video conferencing, telepresence, and how to make your way around the range of available conferencing solutions. We sat down with John to find out what companies should consider to get the most from a video conferencing solution in terms of ROI, TCO and QoS and to get a better idea of the differences between video conferencing solutions, such as HD video conferencing, telepresence and desktop conferencing.
Would you agree that many enterprise organizations are not doing their due diligence in terms of network testing and analysis before they launch a video conferencing suite or new UC technology?
John Bartlett: It is true. Some organizations have this under control and have network architects that understand the problem and do their due diligence, but many do not. The real issue here is that video conferencing is real-time traffic, as is VoIP, so many of the problems are similar to what you'd experience with VoIP -- the difference being that video takes up tremendously more bandwidth.
What kind of due diligence do you recommend in order for a company to launch video conferencing technology?
Bartlett: There's a lot of preparation to be done, and a lot of learning to be done, because real-time traffic behaves very differently in the network and requires the network team to have new tools to understand it and to approach the problems differently.
My experience in dealing with enterprises is that part of the problem is doing the network assessment and figuring out how they need to modify the network. But another part of the problem is learning how to get teams to understand what it means to manage real-time traffic on networks.
Companies need to evaluate carefully whether their networks can support this, and it's best to try to get them help for that from someone who can understand the problem and help them outline the task and figure out what needs to be done in the environment in order to ensure that they can carry real-time traffic and so that they can continue to do that. There's a monitoring task to constantly check the network after they've deployed and make sure it's staying in good shape.
Once a company has taken the necessary steps to understand its network, what's next?
Bartlett: The first thing that I always do when I'm working with enterprises is to assess the demand, and what I mean by that is the amount of bandwidth that will be generated by video conferencing. This requires some understanding of the application, because video conferencing can run on relatively low bandwidth if you're talking about small images on a desktop, and it could be consuming high bandwidth if you're talking about telepresence suites, and there's a very broad range in between. You have to figure out that quality issue, which will determine how much bandwidth is being used per endpoint.
You then have to look at the volume -- how many calls will you support simultaneously in your environment and how does that map to your network topology. We can say this Chicago link is going to need this support -- 15 simultaneous calls -- but the LA link might need to support only five.
The third piece is to understand the bridging architecture of video, because the bridge is this bandwidth hotspot where everything comes together, and that needs to be placed very strategically to make sure that it gets the bandwidth it needs without costing too much money for the bandwidth and the network.
Then, is there a quality of service (QoS) deployment in the network? Do you have the right classifications? Is video conferencing being segregated out as its own class? Which is the right approach? Is that QoS deployed properly all the way across the network, with the bandwidth allocated, so that it gets the bandwidth it needs when video conferencing is running?
What tools or methodologies do you recommend to ensure consistent QoS?
Bartlett: There are two important pieces -- one is bandwidth management. Once we set up the network for video, we establish a certain amount of bandwidth on each link. So let's think about that Chicago link with 15 calls. It's going to need 25 MB of bandwidth for video. That gets programmed right into the network so it will support up to 25 MB of video, but after that, it starts dropping packets. So if the video demand exceeds what we've designed, it will start dropping packets and affect the quality of the Chicago calls.
There's an important piece called bandwidth management, but it is also sometimes called mission control, which ensures that the number of calls or the bandwidth that's created by those calls will not exceed the design spec for your network at any time. That means that in some cases, it will cause busy signals or disallow calls because the bandwidth is being used.
Is it necessary to purchase network monitoring tools to specifically address potential QoS issues with video conferencing?
Bartlett: Network monitoring tools are important. The key difference between the way we monitor our networks today and what you need for video is that you need to understand the behavior of the network path, so you need tools that will look at the behavior of packet streams flowing from one video endpoint in the network through the core and all the way to the other video endpoint and you must look at many of those paths in the network. Most data network monitoring tools look at the devices, the routers and the switches, but don't look at the behavior of the path, so they will often miss problems that cause quality degradation for video.
There are a number of vendors that specifically design tools for voice that work well in the video environment -- NetQoS, Apparent Networks, Brooks Networks, Fluke Networks -- that have some tools designed specifically for real-time applications and that path kind of testing.
The other approach is what's called IP SLA, and that is a functionality that is provided in Cisco routers. If you enable it correctly, you can monitor paths, not quite end-to-end but close, at least router-to-router, to determine what the path behavior is across the network, and that's what's most important in terms of measuring the ability of the network.
What are the primary differences between HD conferencing and telepresence solutions?
Bartlett: HD conferencing is just an extension of standard video conferencing. It tends to be either room-based systems or executive desktop systems with a single screen, and in some cases, two screens (the second being for data shared info from your desktop), but only one screen of images of people on the other side.
Of course, the difference between standard and HD is that HD has high-definition resolution, so you get much clearer images. There is a single codec. The codec is the piece of hardware that is responsible for taking images from the camera and turning them into packets on the network, and the other end is taking the packets off the network and turning them back into images for the screen. So there's a single codec on each end.
The higher-end telepresence systems are mostly codec systems. You sit in front of three or four screens, and there's a codec and a camera for each one, so you get a much broader view of the room.
What are some other differences between HD conferencing systems and telepresence?
Bartlett: A telepresence system is also very carefully designed to give you this immersive, in-the-meeting kind of feel. Another thing they have is stereo sound so that if someone is speaking to you on the left, the sound comes from the left, and if someone is speaking to you on the right, the sound is coming from the right.
Another major difference is you have a dedicated room where the participants face the screen. You're really meeting with the people on the other side of the video, not with each other, whereas very commonly, the HD system will be deployed in a conference room with a long, board-room table with the video system down at the end, so the remote participant feels that he's the king at the end of the table, looking down. You can't necessarily see everyone's face because it's blocked by the people closer to you, so it's not nearly as immersive an experience.
By now, many of us have seen Cisco's TelePresence in action, but for those who haven't, will you please describe the experience?
Bartlett: The telepresence people manage the environment, like the lighting. With Cisco's TelePresence, there's lighting right around the back of the system that shines in your face. The reason they do that is to make sure your face is properly illuminated so that you don't have shadows in your eyes and so that people can really see your face and the camera can pick you up well.
They have carefully placed microphones and also specify the paint colors of the room and even the furniture, so you have this sort of curved desk in front of you and the other participants have a curved desk in front of them, so by matching it up, it looks as if you're sitting around the same table. They have the same color, the room has the same color, so there are lots of psychological tricks to make you think you're in the same room. Of course the screens are showing the people in the far side in life size. They're big screens so that my face would be life size as you face me across the table, which again is trying to convince you that I'm in the room.
Telepresence has that sort of all-encompassing feel to try to solve all of these problems and trick you psychologically to make you feel as if you're in the same room, whereas HD is a smaller version that doesn't include any of those environmental attributes, which largely accounts for the price differences as well.
What are the price differences between a telepresence solution and an HD- conferencing solution?
Bartlett: A big telepresence suite with two to four screens and a room environment like Cisco's is going to be at least $150,000 and up to maybe $500,000 for the room -- to install that and make it go. You can get that HD system for probably $20,000 or $30,000, depending on the options you pick.
What do you think about desktop video conferencing? Is this a viable conferencing solution for companies?
Bartlett: I think there probably are very good applications for desktop video, and part of my goal with this session at Interop is to try to bring some of those out and get people to talk about them.
With desktop conferencing, in almost all cases, you're talking with other desktops, and each image has a single face in it. So what you see of me is my head and shoulders and you don't need nearly as much resolution for a headshot like that, to see what's going on with my face, as you do if I'm sitting on the other side of a conference table some feet away, right? And really, what visual communication is about is reading body language, and much of that is in our faces. So you want to know whether or not I flinch when you say to me, "That's going to cost $75,000." And you'll be able to see that because it's a head shot. The resolution of the newer desktop video system is pretty remarkable. And so, you can in fact get pretty big images on the screen that are quite good, and you can also shrink them down. They have lots of options for how you push them around, and you can format your own screen.
What is the downside to desktop conferencing?
Bartlett: One of my issues with it is that I'm always starved for screen space. One of the things I want to do when I'm collaborating is talk about a spreadsheet or document that I'm working on, and I need to have that up on my screen as well. I think there's a little bit of a fight there when you put the video on the same computer screen as the data that you're working on, and I'm starting to think that I need another screen. I have two now, and I think I need three, and that could get out of hand.
But I think there are some really well-developed applications which will help communications, especially with working teams within a company, or even across companies, between partners, where you're working together on a regular basis and there's little chance of meeting new customers or pitching sales. But the applications will help with an intimate work environment for people working together on projects. I'll be very interested to see what people on my panel and folks in the audience at Interop have to say about it.