December 1998

The Electronic Mosaic: Infrastructure Challenges Posed By Video Applications

BY PHIL KEENAN AND ROGER HECKER

When you compare the amount of bandwidth that data transfer applications consume, you can see that video has a ravenous appetite. This insatiability is not so much due to the amount of bandwidth needed at any given time; rather it is a function of video's need for constant streams of bandwidth for the duration of each application. Many network managers are still unsure about how they will satisfy video conferencing's bandwidth needs. Some become naysayers, hoping to ward off the inevitable, while others nervously submit "rewire the whole building" budgets that have senior management scratching their heads, thinking, "Didn't we just do this?"

Will "the answer" for high bandwidth applications be ATM, Gigabit Ethernet, HDSL, VDSL, or some technology that we haven't heard about yet? One thing is certain: the possibilities for supplying video are many - perhaps too many. As for guessing which solution will be adopted as "the answer," the surest bet is that it won't be just one solution, but a combination of many. Such a combination will demand flexible networking infrastructure equipment that can handle the different networks that exist presently, and that can integrate future networks into the existing hardware. Infrastructure equipment must be designed to accommodate the multiple networks today and tomorrow.

TODAY'S NETWORKS
Today, video generally runs on some sort of dedicated, off-the-LAN network: dedicated lines, ISDN, some dedicated ATM. However, video is beginning to push its way into non-dedicated, on-LAN networks - fast IP backbones, ATM backbones, even some highly segmented Ethernet to the desktop. But, as you probably have heard, most existing LANs don't have troughs big enough for video.

So what about the video that is flying around on these networks now? Setting up a video conference requires more than simply moving data from point A to point B. Each network format carries its own video conferencing standard, and each standard is dependent on the network being used - disallowing, for instance, a video endpoint on an ISDN network from simply calling a video endpoint on an IP network. The call simply can't be connected. The same goes for ATM with either ISDN or IP, or any combination that includes more than one network. The only calls that work are those that are ISDN-to-ISDN, or IP-to-IP, etc.

USING GATEWAYS
Two-way gateways are available that allow, for instance, an ISDN endpoint to call an IP endpoint. These gateways are widely available, are not prohibitively expensive, and give good performance for small video networks. But there is no way of knowing which type of network you will need to video conference with next, so investing only in two-way gateways could be risky.

Even if you know that all of the endpoints in your network will be on either IP or ISDN networks only, what happens when you need to connect to an ATM endpoint outside of your network? You'll buy yet another two-way gateway. And what happens in another three years, when there may be yet another network and standard? When making an investment in video-networking infrastructure equipment, the last road that you want to start down is that winding path of piecemeal solutions: stop-gap measures that satisfy your immediate needs but are not prepared for the long haul.

DEFINING A FLEXIBLE, FUTURE-PROOF, SINGLE UNIT SOLUTION
There are solutions currently available that use many different units: an additional gateway for every network, a different MCU (multipoint control unit) for video processing on each network, multiple PCs for management of each of the units, and modems to communicate between the PCs, servers, MCUs, and gateways. For platforms that were designed to accommodate only single-network communication (such as PBXs for ISDN), these multi-unit methods are the only way to keep up with new technologies. These existing platforms are unable to accommodate new technologies, forcing them to tack-on additional hardware. However, this results in a rather clumsy version of a "solution," and offers merely a makeshift capability.

Alternatively, there are truly "future-proof" platforms: single-unit solutions that provide interconnectivity between all video networks, MCU processing for all types of video streams, and a simple, integrated management interface for all system functions. Such platforms are able to accommodate all current network technologies - ISDN, IP, ATM - and are designed to integrate future network technologies, all within the same unit. Such a solution is designed for the future, ensuring solid investment protection with superior functionality.

Look to meet the following criteria in your video networking equipment:

Universal Transcoding
Each network type uses specific protocols that define the nature of the video and audio streams being sent between participants. The combinations of protocols and standards are legion: for video there is H.261 or H.263, CIF (Common Intermediate Format) or QCIF (Quarter Common Intermediate Format), 7.5 frames per second up to 30 frames per second; for audio there is G.711, G.728, G.722, G.723, or G.729. In a multi-party, multi-network video conference, the number of possible protocol combinations is tremendous, and these combinations can wreak havoc with both the networking equipment and the endpoints. The number of "standard" protocols being deployed is only increasing, making multi-party video conferencing even more complex.

Most video networking equipment forces all endpoints in a video conference to lower their protocols to the lowest common denominator of all participants, lowering the quality of the conference. For example, one of the most common problems for such equipment is to translate between the audio algorithms G.722 (used in high-quality, high-bandwidth ISDN endpoints) and G.723 (used in low-bandwidth, such as 128 Kbps, IP endpoints). With such a combination of protocols, most video networking equipment will force all participants to change their audio transmission protocols to G.711 (the mandatory lowest common denominator for audio algorithms). In a case where the IP endpoints were connected at 128 Kbps, this downgrading of capabilities will force the IP endpoints to use 64 Kbps of their total 128 Kbps for audio (rather than the 6 Kbps that G.723 is designed to use), leaving only 64 Kbps for their video. The resulting conference would be of such poor quality that the participants might wish that they had stuck to a telephone conference. Such situations arise with networking equipment that is unable to connect IP and ISDN protocols using their own terms.

Interoperability between multiple participants from multiple networks, while maintaining the highest video and audio quality for each participant (rather than bringing all participants to the lowest quality), requires advanced equipment. Transcoding of all existing protocols and characteristics for every party from every network is a prerequisite - after all, if the equipment has difficulty handling today's technologies, it certainly will not be ready for the technologies of tomorrow.

Robust, reliable telecom architecture
PC architectures are appropriate for PC users; they are less acceptable in mission-critical environments, where significant revenues depend on the ability of the network equipment to keep working - even under the heaviest loads. PC architectures prohibit quick access to modules, rendering maintenance procedures tedious and usually requiring a system power-down for any configuration modifications. While CompactPCI represents an attempt to bring telco reliability to PC architecture, in our opinion, at this time this cannot be accomplished in a reliable manner.

Telco-grade equipment, however, is designed to continue functioning under the most demanding conditions. Telco-grade equipment is characterized by front-accessible, hot-swappable modules, redundant components, resource reallocation in the event of failure and advanced diagnostics - all available while the system is up and running. You will get a lot more use out of equipment that is designed for maximum access and minimum down-time.

Rich video-processing feature set
One of the goals of video communication is to make remote meetings more lifelike, more like face-to-face communication. A feature called continuous presence allows participants to see more than one other participant on the screen, to make each video conference seem more real. When evaluating continuous presence features, make sure to check for the following characteristics:

• Number of different layouts (should offer more than just quads and full screen).
• The ability to switch smoothly between layouts during a conference.
• Maximum number of participants in a continuous presence conference .
• Maximum number of simultaneous continuous presence conferences.
• The ability to set each window in the layout as either voice-activated or fixed to a particular site.
• The ability to use T.120 data collaboration in conjunction with continuous presence.

High capacity
Over the next few years, your network is going to grow, probably significantly. Low capacity solutions, simply put, are poor investments. A platform that can handle high capacities today, and that has a clear, executable development path to accommodate capacities in the future, should be a mandatory characteristic of any video networking equipment.

The true measure of system capacity is not only the maximum number of total participants in the system, but the maximum number of feature-rich conferences. Features such as continuous presence severely limit the number of simultaneous conferences that some systems can offer. For example, a system that can normally handle 30 participants in 10 different simultaneous conferences may be able to handle those 30 participants in only 2 simultaneous continuous presence conferences. Be wary of the limitations of some equipment.

PREPARING FOR THE FUTURE, TODAY
At this crossroads in telecommunications technologies, you must choose which video networking infrastructure equipment will take you beyond the next networking fad. The solid investment decision is the one that thinks beyond next year and the year after that. The pace at which your organization adapts to developments in network technology will be determined by how powerful, flexible, and adaptable your equipment is.

Phil Keenan is the senior VP of worldwide sales and marketing and Roger Hecker is the marketing services manager at Accord Video Telecommunications. Accord develops next-generation network solutions that are revolutionizing the conferencing industry. The company markets and supports multipoint gateway and control systems for the multimedia conferencing and collaboration industry. For more information, please contact the company at 770-641-4400, or visit their Web site at www.accord.co.il.