Significant strides have been made over the last year in getting the world to finally understand that DSL is just a modem technology. High speed, always on, but still just two modems and a wire. Providers are now recognizing the services traveling over DSL are infinitely more interesting, and orders of magnitude more profitable, than DSL itself.
Fundamentally, DSL has evolved from a high-speed Internet technology to a high-speed multi-service access technology capable of simultaneously delivering a broad set of dissimilar services over the same existing copper wire.
Adding Voice
One of the first new services to be deployed over DSL, alongside Internet access, is voice, which makes sense since voice services still consume a much larger percentage of small-business communications expenditures than data services. Any profit-driven CLEC will wisely address the largest revenue targets first.
Voice over DSL, or VoDSL, actually refers to a set of technologies that allows multiple voice lines to share a single DSL loop while still leaving adequate bandwidth to support multiple high-speed data services. This sharing is accomplished through the use of a multiplexing protocol, such as ATM, that allows bandwidth to be consumed dynamically while still providing each service the QoS parameters it expects. Standards development and vendor products have evolved to the point where ATM adaptation layer 2 (AAL 2) compressed voice with dynamic bandwidth allocation and silence suppression/comfort noise insertion can be supported in a multi-vendor environment. The number of voice lines that can be simultaneously supported is limited more by bandwidth considerations than technology. Nevertheless, two to 48 voice channels have been shown to be practical with 12 to 16 being the approximate mean of the demand bell curve.
While VoDSL represents an enormous advance in the state of DSL technology, it is only a first step. To understand why, it is useful to take a step back and assess what is really happening.
Technology aside, most current implementations of VoDSL provide static connections between Class 5 end-office voice switches and analog telephones, key systems or PBXs located at the customer premises. The service VoDSL enables would more properly be labeled "local dial tone." Every single residential and business subscriber in the United States consumes local dial tone, but there are other voice services that are equally important and potentially much more profitable. Long-distance voice and voice VPNs are examples of additional voice services deployable over DSL. However, these services require extensions to current VoDSL technology.
This does not mean today's VoDSL products do not support, at least indirectly, long-distance telephony. The implementations are actually transparent to the Class 5 switch and, therefore, all voice services currently provided by the Class 5 switch continue to be available to the subscriber. In addition to custom local area signaling service (CLASS) features, this includes routing long-distance calls to a Class 3/4 switch operated by a long-distance provider.
However, there are powerful economic incentives to the access provider associated with bypassing the local Class 5 and delivering long- distance calls directly to the long-distance provider. This is especially true of CLECs lacking their own Class 5 infrastructure, as well as long- distance providers aiming to terminate long-distance calls themselves rather than paying a LEC to do so.
Supporting long distance over DSL, or, more accurately, local bypass over DSL, requires certain extensions to current VoDSL technology. The goal of most VoDSL products is actually transparent local loop emulation. Supporting long-distance voice requires the VoDSL infrastructure to evaluate the call destination and determine the most appropriate switch to which to connect the subscriber: Class 5 or Class 3/4. This, in turn, points toward two new technologies that must be added to VoDSL to make it applicable to a dynamic local/long-distance telephony environment: switched virtual circuits (SVCs) and multimedia gateway control protocol (MGCP).
ATM SVCs are required to optimally route voice calls without requiring an unscalable number of permanent virtual circuits (PVCs) to be preconfigured. VoDSL network infrastructure, especially in networks operated by nationwide CLECs and IXCs, is likely to scale to hundreds of thousands, if not millions, of customer premises IADs. Networks of this size must rely on switched connections to scale efficiently.
It is obviously impractical (and also unscalable and unmanageable) to expect each IAD to contain a call- routing database to determine optimal call routing. External, logically centralized, physically distributed databases are needed to provide IADs with instructions for call routing. These databases are represented by call agents within the network and are accessible using protocols such as MGCP. The call agents, in conjunction with the packet-based VoDSL access network, form the basis of next-generation distributed multimedia switching architecture.
It is also possible (for larger businesses, likely) that a significant subset of call volume is not local or long distance, but rather between corporate locations--a branch office calling a corporate PBX, for example. Here again, there are strong economic incentives for the service provider and the subscriber to route the call over the access provider's ATM backbone (or even a third-party network provider) rather than through the PSTN. Like long distance, this requires the VoDSL IAD to evaluate the call destination and make the appropriate routing decision. This type of "on net" call is sometimes called a voice VPN. Properly structured, it allows service provider customers to dial three- or four-digit extensions and connect with other corporate personnel.
Diagram: Next-Generation Voice
These voice services--local dial tone, long- distance voice and voice VPN--are merely examples of products the service provider is able to market to business subscribers using more advanced VoDSL technology. A typical example might be $18 per month for local dial tone, 8 cents per minute for long distance and 6 cents for a voice VPN. Yet implementing these services requires significant extensions to VoDSL. In fact, these extensions--SVCs and MGCP--are really not tied in any way to DSL and can, therefore, be implemented over T1, NxT1 and T3 access facilities as well.
Adding Frame Relay
After voice, the next logical product to add to the DSL service bundle is frame relay. Frame relay is interesting because it is not actually a service at all; it is a WAN access protocol. The service enabled by frame relay is actually a Layer 2 VPN or a transparent LAN service. Each of these connects corporate locations using frame relay virtual circuits and RFC 1490 between frame relay access devices, LAN switches or routers. Nevertheless, frame relay has become extremely popular among business subscribers with more than a single business location. It is radically less expensive than leased-line networks and exhibits better quality and greater security than Layer 3 VPNs over the Internet.
The economic benefit of moving frame relay delivery from conventional leased (from the ILEC) FT1/T1 facilities to DSL is enormous. Current frame relay deployment methods call for the service provider to lease access facilities costing several hundred to a few thousand dollars per month. Moving to DSL reduces the access cost to $10-$20 per month. Even after factoring in collocation capital costs and backhaul charges, the numbers are overwhelmingly in favor of DSL.
CLECs can offer frame relay over DSL in one of two ways, closely paralleling VoDSL deployment alternatives: by reselling existing frame services or by selling native frame services over the CLEC's own backbone. Neither case requires the CLEC to operate a frame relay infrastructure since the frame relay traffic emanating from the CPE (e.g., a router) is immediately converted into ATM using FRF.5/FRF.8 and delivered to its destination as ATM cells. The IAD must, of course, map frame relay-specific QoS parameters into ATM to preserve service quality across the entire network.
Kevin Walsh is vice president of marketing at Accelerated Networks Inc. (www.acceleratednetworks.com). He can be reached at (805) 553-9680.
