Too HOT to Handle?

Data is hot. Everyone knows that. Everyone wants a piece of the Internet action. What no one seems to have realized yet is that data, or Internet access, could become too hot for the POTS network to handle. The current telephone network is facing an Internet meltdown. That is a scary thought because the POTS network carries the voice traffic that so many people count on, whether it is on Mother's Day, during a power outage or in a critical situation. The POTS network is also the transport that enables the explosive growth in the Internet.

Telcos have virtually 100% market share for local Internet data transport, which means that today the POTS network is something of an Infobahn. It also means that the burden of supporting the hyper-growth of the Internet falls directly on loops and switches that were designed and built for voice traffic. Reinforcing the current POTS network and switching fabric to carry expanding Internet traffic could cost more than $18 billion, or $1200 per line. Upgrading the network using a technology called switched digital video plus (SDV Plus) can achieve the same goal-for about half the cost.

Internet Meltdown The Internet meltdown threat arises from two factors. The first factor is the difference in call hold times between Internet calls and voice calls. The average voice call lasts about four minutes. The average on-line service (OLS) connection grew from 19 minutes in 1992 to over 33 minutes in 1994. Over 20% of OLS connections lasted more than 60 minutes in 1994. It is expected that 1995 will show continued increases in both average time and the portion of users that stay connected more than one hour.

The second factor is that data calls do not replace voice calls. They are separate activities that often occur simultaneously within a household. This factor is fueling the growth in second lines. It is this combination that is driving the POTS network toward meltdown.

Traffic engineers size switches based on a usage assumption of about nine minutes per line during peak hours. What would it take to double that load? The answer is just 15% penetration for Internet usage. If just 15% of households went on-line to the Internet at one time and had a call-hold time of one hour, it would double the current capacity demand. This has the same effect as 100% of households making a nine-minute call in that same hour. With on-line data services, a relatively small user group can stress the network in ways never before experienced.

What would doubling the capacity demand mean to network costs? The current U.S. telephony switch investment totals $18 billion in switch fabric and $12 billion for ports deployed to serve 100 million U.S. households. Because the costs for network capacity increases are approximately linear, the cost of doubling capacity to support these Internet users will be an additional $18 billion in switch fabric and $12 billion in new switch ports. Reinforcing the local copper loops will add billions of dollars more. Reinforcing the POTS network fast enough to handle the hyper-growth of the Internet may simply be impossible.

How likely is this meltdown scenario? In 1995, PC penetration reached 33% of households and is rising rapidly each year. On-line services have reached 7% penetration. Sixteen million users access the Internet in the US., and with Internet usage growing at 60% annually, 26 million users could be on-line by the end of 1997. The effects of longer hold times can already be seen in the network. In its 1995 annual report, Nynex indicated that it experienced 3% line growth and 8% utilization growth. Switch vendors report pressure from customers to increase their utilization assumptions. We aren't there yet, but the demand curves are headed toward a network in which data demand overpowers traditional voice service.

What does this hyper-growth in Internet and data demand mean for the telephone company's business case? Disaster. Virtually all data calls are local connections to an OLS point of presence billed as part of local service. Revenue growth for the telco from data service connections will be minimal, while the costs of reinforcing the network to carry data are likely to be high. Even worse, as voice over the Internet becomes a reality, long-distance access fees will be lost.

And even if telcos make the investments to reinforce their networks, they could eventually lose the data delivery business to cable TV companies that will offer ubiquitous high-speed data connections within the next 10 years. To add insult to injury, cable operators plan to use the POTS network as a return channel, again resulting in long hold times and increased capacity demand with only a minimal slice of the revenue going to the telcos. Data over the POTS network is a recipe for the railroad-ization of the telephone industry, hauling information coal for pennies per ton.

Data Service Requirements The answer to the data meltdown scenario goes beyond throwing money at the network. Rather than simply reinforcing the current POTS network, the telephone industry must upgrade the local loop in a way that preserves the strengths of the POTS network while meeting the special requirements of data service.

One of the benefits that consumers get from using the POTS network to access the Internet is that they can dial in from anywhere, and they do not have to deal with service degradation due to congestion. These benefits exist because the POTS network is a switched network that dedicates bandwidth to each home and assures information providers that their services only reach the intended party. A switched network is secure against service piracy. The switched aspect of the network must remain in any upgrade scenario.

Accommodating data requires a few changes, however. The first is that data service must be separated from voice service. Voice telephone service is the crown jewel of the communications industry and cannot be compromised. In addition, separating voice and data services makes sense because of differing standards: Voice switching is evolving toward TR-303 switching standards and data switching is evolving toward asynchronous transfer mode standards. Because using a single standard would add unnecessary translation costs, it benefits neither service.

In addition, the bandwidth requirements of the two services are very different. Voice is a low-bandwidth, symmetrical service that cannot accommodate transmission delays, and data-especially data that includes video-requires bandwidth that is one or two orders of magnitude higher than voice.

Internet data traffic also tends to be asymmetrical, with more traffic going downstream than upstream. Delay or latency requirements are variable for data. Some data services can tolerate delay, others cannot.

Although voice service should be segregated from data service, both data streams should be carried on a single integrated local loop network to keep operational costs low. Integration also allows the services to be more easily sold on a bundled basis from a common platform.

SDV Plus Networks The solution for telcos to accommodate the hyper-growth scenario for data services, while maintaining reliability, quality and cost standards for voice, is SDV Plus. Originally developed to deliver video services, SDV was the first asymmetrical digital service. Its new name, SDV Plus, recognizes the ability of this technology to deliver data as well as video services.

An SDV Plus network is a fiber-to-the-curb (FTTC) system that incorporates ATM edge switching and telephony transport in a single integrated network to deliver the whole service bundle. It delivers data and video services via ATM and simultaneously provides TR-303 or TR-008 standard telephony service. An SDV Plus system can be deployed for about $600 to $800 per home including the ATM switch, fiber, electronics and installation-about half of what it would cost to reinforce the POTS network to avert an Internet meltdown. The integrated architecture of SDV Plus also achieves significant operational savings once it is installed.

As shown in Figure 1, telephony enters the SDV Plus system in DS-1 through a standard next generation digital loop carrier switch interface, either TR-303 or TR-008. The switch interface element also provides the time slot interchanger for provisioning and grooming DS-0 traffic and provides the interface to the operations support systems (OSSs), which allows telcos to administer their POTS traffic with existing OSSs. The DS-1s that carry the telephony traffic are then passed to the ATM interface shelf where they are multiplexed together with the ATM data traffic and sent to the optical network unit (ONU). At the ONU, the DS-1s are demultiplexed and passed to a shelf containing standard telephony channel units. The channel units are the same ones used in existing DLCs, again ensuring that no new technology is needed to support the existing services. The result is an evolution, rather than a revolution, of telephony service.

The SDV Plus architecture allows the data services to move from the telephony components to the ATM video and data components (although customers who like their telephony modems can still use them if desired). The ATM traffic arrives from Internet service providers via the existing Sonet network. Point-to-point traffic is routed through a centralized ATM switch. Point-to-multipoint services do not require use of the centralized ATM switch; the edge switching in the SDV Plus element can provide broadcast switching. The fiber-star, ATM transport provides dedicated bandwidth so that as data traffic increases, the system performance does not degrade.

Finally, the SDV Plus architecture offers much higher bandwidth, 52 Mb/s downstream to each home and 1.62 Mb/s upstream from each home. The result is that the long hold times and high-bandwidth demands do not burden the voice system. ATM standards and networks are designed to be scalable as data traffic grows in usage and speed so that this network can keep up with the Internet.

Switching in the POTS network offered the security of knowing that the data was delivered directly to one house. ATM switching in the SDV Plus network offers the same benefit. The SDV Plus network will not deliver data to a home unless the delivery has been authorized by the video or data OSS. Data and video piracy, which costs the cable TV industry 14% of its revenue, is not possible with SDV Plus.

Finally, the SDV Plus architecture integrates the two services onto a single transport platform. The network operator only needs to maintain one network. Maintenance costs can be as much as 50% to 60% lower than the current method of operation. When compared with maintaining separate voice and data/video networks, costs may be as much as 80% to 90% lower. When lower maintenance costs are combined with increased intelligence in the network for proactive performance monitoring and the reliability of fiber, trouble reports can be reduced dramatically, thereby increasing customer satisfaction. Several network operators believe that operational savings can pay for the network upgrade to SDV Plus.

Long hold times and increased bandwidth requirements threaten to overwhelm the POTS network. Spending money on reinforcing it is doomed to be a bad investment decision. Now is the time to plan for the movement of Internet and data traffic off the POTS network and onto a segregated but integrated high-speed network. The SDV Plus architecture offers an alternative that builds on telephone network strengths and protects its high standard of reliability for voice service.

Don McCullough is Product Line Manager at BroadBand Technologies Inc., Research Triangle Park, N.C.