Will the VoDSL heyday hit payday?

Providers have made the investment in DSL technology - now it's time to reap the rewards

Recently, data competitive local exchange carriers have been held as examples of everything that's wrong with DSL. Their stock prices are plummeting, and according to The Strategis Group, their average gross margin in 1999 was -71%. Yes, negative percent.

Yet end user demand for DSL service is booming at a growth rate of 50% from quarter to quarter, according to Cahners In-Stat Group. Service providers have invested heavily in this potential. They're just waiting for payday.

But many factors prevent DSL from reaching its potential. Some of the problems, such as loop quality and order management, are being addressed by solutions such as automated loop pre-qualification systems and DSL-specific back offices.

Even more issues arise when service providers want to leverage DSL infrastructures to deliver voice services, as is the case with voice over DSL.

Integrating services this way is done for various reasons. First, voice services pay the bills, but data services win markets. Second, packet networks use bandwidth more efficiently than circuit-switched networks.

The voice-over-DSL opportunity is as promising as DSL itself. Voice over DSL is expected to reach 2.9 million lines by 2002, 22% of the total voice-over-packet market, according to US Bancorp Piper Jaffray.

But integrated broadband services such as voice over DSL present challenges that require an entirely new way of looking at and dealing with the access network, that last mile of infrastructure that DSL is meant to broadband-enable.

Resolving these problems will do more than help just voice over DSL see payday. A resolution is critical for service providers to see payday from their integrated service offerings.

Big, hairy hang-ups Big, hairy hang-ups are network-based problems that have a direct effect on service providers' bottom lines and their ability to execute innovative business plans.

Excessive provisioning. Many network elements must be "touched," and many disparate circuits created, to enable a new service or subscriber

Slow turn-up. There is a long lag before services can be deployed and revenue gained

High cost. Service providers must consider expenses of network provisioning, of operations and of maintenance

Quality challenges. Service providers must guarantee a high level of quality, which is now a table stake - something the provider needs in the product to make it work - and must differentiate based on quality of service (QOS) for integrated service offerings

Each of these problems has roots - and solutions - in the access network.

Any competitive carrier serving residential and small and medium-sized business customers - the largest markets for DSL services - must do so over an access network. The access network connects an enterprise or residence to the edge of a carrier or service provider network.

Current network architectures treat the access network as a collection of generic pipes that must be bulked up with numerous pieces of equipment, including switches, routers and gateways, and fortified with fatter pipes and larger fabrics to decrease the probability of congestion and deliver integrated broadband services. In essence, complexity is increased to move more bits.

This creates an expensive system and an intricate one - one that is susceptible to complications. More problematic is that these approaches add to the severity of the big, hairy hang-ups.

Provisioning Of all the network hang-ups, provisioning is certainly the hairiest. It's expensive. It's time consuming. It can hold up business plans and prevent broadband service providers from seeing real profit. And, it's the toughest problem to reform because it's rooted in the architecture of the access network.

In today's circuit-switched network, a customer-located integrated access device (IAD) is connected with a single permanent virtual circuit (PVC) to each element along a service's path. In addition, PVCs are provisioned from hop-to-hop (between each element) to pre-define and guarantee the route and quality of a particular service.

In a typical DSL configuration, only the elements required to deliver three basic services to a single end user are shown. The services include voice service, data service and virtual private networks.

To connect all 10 required elements to deliver the three services, 12 PVCs are needed (PVCs are not required from the Class 5 switch to the gateway or public network).

Provisioning for tens of thousands of customers becomes excessive. And provisioning may be increased by errors - TeleChoice reports that "errors in service providers' provisioning process prevent DSL from successfully being deployed in the first go-round 85% of the time."

In this way, it becomes impossible to turn up services with the speed and margins that service providers require to be successful.

Slow turn-up Obviously, the provisioning burden has an impact on the speed at which services can be turned up. And by nature, DSL introduces additional lag. Loop qualification for DSL service can take months, and the results are far from accurate.

Not only does DSL have a deployment failure rate as high as 85%, a typical DSL service requires truck rolls.

DSL is being turned up as fast as service providers can manage it. But service providers also are limited by the availability of network equipment, which vendors are producing as quickly as possible. And the equipment vendors, in turn, are limited by the availability of chips - chip vendors are oversold, and silicon is becoming an endangered species.

Service providers, vendors and end users are racing toward a single technology to deliver their services. And the result isn't speed. It's a log jam.

Cost and quality Service providers are caught in a Catch-22: The most promising business model is also the least profitable. Voice services still pay the bills, but a service provider must offer data and integrated services to win the market.

The cost of purchasing and maintaining the pieces of the access network required to deliver integrated services in today's access network architecture is prohibitive. RHK reports that service providers' total capital expenditures continue to rise at about 21% per year.

And, retail cost of DSL (and other data) services has been falling while its deployment costs remain high.

Then there's QOS, which isn't what it used to be. Today, quality itself is a service. End users still expect traditional table stakes such as toll-quality voice 99.999% of the time, and more recent table stakes such as "always-on" Internet access and high-speed download capability.

But now quality is an increasingly crucial point of differentiation for service providers. When an end user can choose the level of quality for a particular service on demand - and a service provider can bill for that quality - then quality is the service.

But things are getting complicated. Voice over DSL is one example of a service delivery mechanism that packetizes voice for increased line usage. If voice packets are not treated properly - if they are subject to high jitter, delayed or randomly discarded in the way that data packets are - then service quality is affected, and the result is simply unacceptable: choppy video, audio clicks and pops.

To make true QOS possible in this environment, the network has to prioritize and schedule both voice and data traffic. But today's access network, with its disparateness, complexity and inherent cost, can't deliver the goods. To compensate, service providers typically over-provision the network, sacrificing usage and efficiency for peace of mind.

Overcoming the hang-ups None of these problems is solved by adding another box or function to the network. None can be eased by deploying bigger fabrics or laying larger pipes. In fact, those kinds of "Band-Aid" and "steroid" solutions actually exacerbate problems with additional complexity and cost.

To address each of these hang-ups, a new access network architecture is required - one that leverages the current infrastructure to deploy integrated services. The answer is a services-aware access architecture, one that places key intelligence at the "binding points" of the network.

"Book-ending" the access network in this way can give a service provider greater control over that crucial part of the network it relies on to do business. And it can provide converged functionality at key points in the network, reducing disparate elements, and therefore, reducing provisioning.

Furthermore, the provisioning between the two binding points - the IAD and the multiservices switch - is constant. A traditional access network must be provisioned each time a subscriber or service is added because each subscriber requires its own pipe and path to specific services. In a services-aware architecture, however, the network is only re-provisioned when a new IAD is added. This substantially reduces the required instances of provisioning and makes scalability incremental by port or IAD.

In a services-aware architecture, to deliver the same three services to a single subscriber as shown in Figure 2, just six elements and six PVCs are required. Only a single PVC is required for data services, and a second PVC is used for voice. At the headend of the network, the reduction in disparate elements further reduces the PVCs required.

And because the IAD is effectively connected to all service networks via the multiservices switch, subscribers can select multiple services and destination networks on demand.

The effect of this on the speed of service deployment and cost is significant. New subscribers and service networks are quickly turned up because provisioning is minimized and customer-located intelligence quickly manages new subscribers and service connections.

Operational cost is minimized along with provisioning, as is the cost of deployment. A single element is far less expensive to purchase and maintain than many elements. And small but intelligent customer premises devices are more scalable and cost-effective than behemoths in the core.

In a services-aware network architecture, a single intelligent IAD manages integrated traffic for many end users. Thus, numerous subscribers can be served with the deployment of one IAD, effectively slashing the need for truck rolls to a fraction per subscriber.

A services-aware network also supports the quality required by today's applications. By providing total control over the traffic between both edges of the access network, service providers can guarantee bit rates, dynamically allocate bandwidth to users on demand and control the QOS algorithms between the binding points.

With greater intelligence located at the customer premises, a services-aware network can enable end users to self-provision their services in real time, on-demand and pay-per-use. In this way, service providers have the power to pre-engineer services, while the end user is granted the power to dynamically select these services as desired.

Driven by end-user demand, the formula for service will soon include three equal parts: application content, QOS and end user choice. Service providers must anticipate and innovate on each of these three counts - to turn big, hairy problems into audacious goals.

To do so, they need everything in place to support these services and to foster innovation in their offerings - from their back-office systems to their networks. They need speed, efficiency and profit.

In short, they need a simple, intelligent network architecture that lets them rapidly deploy services and make business decisions unhampered by infrastructure. That's a services-aware network.