Implementing On-Demand Video Services Using DSL Technology.

In the early 1990s, cable companies explored the idea of offering video on demand services (VoD) to consumers. While the concept was compelling, the logistics were overwhelming, the technology was not quite ready for mass markets, and the idea did not take off. But you can't keep a good idea down for long. A number of Hollywood studios are now exploring ways to offer feature films and other entertainment on demand over the Internet, begging a key question: How can Hollywood's on-demand "content" be cost-effectively piped into customers' homes?

Because of the distances involved, the technology puzzle has yet to be solved in terms of delivering on-demand video to customers dispersed across single-family residential neighborhoods. But a solution is now available to bring VoD to customers in apartment buildings, condominiums, and hotels: Digital Subscriber Line (DSL) broadband-access technology.

DSL And MDU 

The multi-dwelling unit (MDU) residential market is fertile ground for broadband technology providers. Nationwide, there are nearly 22 million residential units in MDU buildings. About 58 percent of them are in buildings with 10 or more units; 37 percent are in buildings with 20 or more units. Broadband providers have barely begun to tap this market.

To date, broadband providers' primary value proposition for MDU customers has been high-speed Internet access. According to an October 2000 research report by the Strategis Group, by the end of 2000, MDU market penetration for high-speed Internet service totaled approximately 1.2 percent for all providers, including DSL, cable, and wireless. Strategis projects that high-speed Internet service will reach about 3 percent of the MDU market in 2001. In the race to serve this market, DSL providers--who can leverage MDU buildings' installed copper phone wiring to quickly deploy and deliver broadband services--can make their offerings more compelling by adding on-demand video to their portfolios.

Video on demand is very appealing to prospective broadband customers. The Strategis Group found that more than half the people it surveyed on the question expressed interest in receiving VoD services. Nearly a quarter said they were either extremely interested or very interested. It stands to reason: For the end-user, video-on-demand service is comparable to having a continuously updated, large video library in the den or living room.

DSL access solutions enabling cost-effective broadband-service delivery to MDU customers are converging with maturing technologies and improving economics elsewhere in the network to make video on demand a cost-effective service offering for DSL providers. Video servers, which are now offered as off-the-shelf PCs with large disk drives or RAID arrays and Ethernet cards, run special video-streaming software, and offer a low-cost video delivery platform. MPEG2-based video-streaming technology now enables delivery of up to DVD-quality video to customers' TVs, via set-top boxes. The cost of set-top boxes is coming down, thanks to falling prices for processors and other digital components, propelled in turn by the economies of scale associated with increased production of set-top boxes for digital cable TV.

Both Internet traffic and streaming-video traffic are transported in Internet Protocol (IP) packets. TP packet headers are loaded with information that identifies the traffic's precise nature, including its source, its destination, and its key service parameters. The heart of cost-effective bundled Internet access and on-demand video service over DSL is an access solution that can see these packets and take advantage of the information they contain--a DSL access concentrator with IP service intelligence. IP-optimized DSL concentrators' ability to see IP packets enables them to distinguish between different kinds of traffic on a single circuit, aggregate different traffic types into service classes, and distinguish between traffic from different subscribers. This capability pays significant dividends for providers in operational cost efficiencies and savings on capital investment.

ATM Not The Best Solution

The prevalence of Asynchronous Transfer Mode (ATM) technology in the network core has led many broadband providers to deploy ATM-based DSL Access Multiplexers (DSLAMs) at the network's edge. But ATM access solutions are less efficient and more costly to support than IP-optimized solutions. ATM-based DSLAMs segment all traffic into homogenous 53-byte cells. This adds overhead and reduces data throughput, and prevents the DSLAM from seeing IP packets. The valuable information in the packet headers is lost to ATM-based DSLAMs, and they're unable to make decisions based on this information. This means that ATM DSLAMs must transport each customer's traffic on separate permanent virtual circuits (PVCs), and each class of traffic for each customer on separate PVCs. A provider delivering bundled Internet and on-demand video service to 100 end-users across an ATM-based DSLAM therefore has to provision 200 PVCs to serve these customers.

Because IP-optimized DSL concentrators keep packets intact, they can act on packet-header information to make intelligent forwarding and queuing decisions. Their ability to see this information enables them to distinguish between different kinds of traffic on a single circuit, aggregate different traffic types into service classes, and distinguish between different subscribers' traffic.

IP-intelligent concentrators can use packet information to aggregate similar traffic from different subscribers onto shared virtual circuits to and from the network backbone, and de-multiplex this traffic for forwarding down the correct subscriber's line. This means that a provider with an IP-optimized concentrator can deliver best-effort Internet access and video on demand to 100 endusers with two virtual circuits, compared to the 200 PVCs an ATM-standardized provider must provision to deliver the same services to the same number of customers.

Once the IP-aware concentrator aggregates Internet traffic and video traffic into separate service classes, it can treat each traffic type differently, prioritizing, queuing, and forwarding each class' traffic separately for each user. Combining the ability to act on packet information with queuing and buffering techniques similar to ATM-based systems, and traffic parameters on aggregated ATM circuits, IP-intelligent concentrators deliver the service quality each traffic type requires, for every user.

The IP-optimized DSL concentrator's ability to see and direct IP packets in these ways yields important advantages for broadband providers in the MDU market: It enables them to aggregate customer Internet traffic onto a single permanent virtual circuit through the network backbone, and into the POP, dramatically reducing service-provisioning time and costs. It also enables providers to deliver video on demand in a simple manner, without inundating their networks with virtual circuits, or investing in separate routers or switches to handle video streams. Because IP-optimized DSL concentrators aggregate Internet traffic separately from video traffic, providers can deploy them in much simpler networking configurations than ATM-based DSLAMs, and with less investment in upstream supporting equipment, including ATM switches and subscriber management systems.

In short, the IP-optimized DSL concentrator combines the power of IP intelligence with a little help from ATM's quality-of-service capabilities to offer a simple, elegant, all-in-one solution for provisioning Internet service and on-demand video service in MDU buildings.

The MDU Market

Multi-dwelling unit buildings are ideally suited for bundled Internet and on-demand video service, because all customers in an MDU building are close enough to the DSL concentrator--within 4,000 to 6,000 feet--to be served at download speeds as high as 7Mbps over Asynchronous DSL (ADSL) customer premise equipment (CPE). With VHS-quality video clocking in at 1.5 Mbps, ADSL technology delivers speed to spare for broadband providers in MDU buildings. Moreover, ADSL technology enables providers to achieve these speeds over existing plain old telephone service (POTS) copper wiring that is available in all residential buildings.

To offer bundled video on demand and high-speed Internet access to MDU customers, providers co-locate a video server with an IP-optimized DSL concentrator in an MDU building's basement wiring closet (see Fig). The concentrator interoperates with video-streaming software in the server and ADSL CPE to deliver video streams to customers' TVs via set-top boxes, and concurrent, high-speed Internet service to customers' computers via an Ethernet connection. Providers can store hundreds of films on the video server, and continuously update their film libraries by downloading new titles from the Internet.

High-speed Internet access is always delivered on a "best-effort" basis, but where VoD is concerned, nothing less than guaranteed quality will do. Three factors affect broadband providers' ability to deliver simultaneous streams of VHS-and-better-quality on-demand video to customer TV sets: speed, available bandwidth, and proper queuing and buffering techniques.

Customers' close proximity to the concentrator solves the speed issue in the MDU environment. Bandwidth issues are easily resolved by connecting the video server to a co-located IP-optimized DSL concentrator through two 100Mbps Ethernet ports, enabling the provider to secure 200Mb of bandwidth for video--enough to ensure simultaneous delivery of VHS-and-better-quality video to scores of subscribers.

Video-streaming software and decoders cannot tolerate a connection that is too "bursty" without compromising video quality. A smooth, reliable bandwidth stream is essential for high-quality video. IP-optimized DSL concentrators employ queuing and buffering techniques to ensure that video gets the bandwidth it needs while end-users simultaneously download large files--such as MP3 audio, graphics-rich and interactive Web pages, and e-mail with attachments--from the Internet. The concentrators place video in individual high-priority queues for transmission to each end-user, and downloaded Internet traffic in separate best-effort queues, ensuring that video gets the service quality it requires.

For broadband providers, bundled on-demand video and high-speed Internet service is a promising business model. And the MDU market is an ideal space to execute it. In fact, where execution is concerned, IP-optimized DSL solutions--which cost-efficiently transform humble copper pairs into robust broadband pipes--are the best option for providers and end-users alike.

Robert Beliveau is product marketing manager at Copper Mountain Networks (Palo Alto, CA).