BBH Report Home Page
September 15, 2004 Provided by System Dynamics Inc.

DSL Forum -- New Specs to Beat Cable

A July headline in Variety.com read “Telcos beat cable in broadband growth”. For the first time, North American DSL providers added more net residential high-speed data subscribers than cable operators, who had held the lead since the initial broadband rollouts.

The cable operators got out in front by embracing broadband as a new revenue opportunity, using CableLabs to create a standard set of buying specifications (DOCSIS) which drastically cut the costs of cable modems, and making broadband services available to most of their subscribers. The telcos were slow off the mark, each operated independently with different buying specifications, and were only able to reach customers fairly close to central offices. As a result, US cable operators captured about 2/3 of the broadband market, with the telcos getting about 1/3.

Recently the tide has turned as telcos fought back. The initial basis of competition was obvious: the telcos cut their prices. Cable generally did not join in the price-war game, but instead started using their plant capabilities to raise broadband speeds. New advances in DSL technologies allowed telcos to start raising their speeds as well.

After competition has gone thru the pricing and speed games, the next logical battleground becomes the fight over which is “better”. The DSL Forum has recently published new specs designed to change the rules of the game.

We've talked with people in the DSL Forum over the years, but until now we have not looked closely at their reports. With titles like "Recommendation for Physical Layer of ADSLs with a Splitter", they were very low-level specifications for the underlying DSL infrastructure.

In June, we received a press release from DSL Forum about several new specifications which seemed much more about home networking than DSL. That led to a telephone interview with Michael Brusca, Vice-President, Strategy, DSL Forum Board of Directors; and Distinguished Member of Technical Staff:DSL and Broadband at Verizon Communications. We also talked with Tom Anschutz of BellSouth Telecommunications, an active participant in DSL Forum and the editor of one of the key specifications.

In the interview, Michael told us about the new family of Technical Reports published by DSL Forum over the past year. After the interview, we took the time to download and read through the reports we had discussed. Taken as a whole, they represent a dramatic change in the role of DSL Forum -- from engineers defining the nitty-gritty of wiring and bits, to strategists defining what may well be the future course of the telephone companies.

We've been around the telecommunications industry for some time, and we remember reading reports like these published years ago by the old Bell Labs when AT&T was the US telephony industry (how many of our readers remember "BSPs"?). It certainly appears that DSL Forum is filling -- on a more-or-less volunteer basis -- the vacuum left by the inability of BellCore to play the role envisioned for it when AT&T broke up and BellCore took on the coordination responsibility for the network.

Michael walked us through the whole family of Technical Reports. They envision a multi-services network connecting multiple users in the home, working with multiple devices over a home network, to multiple services provided by a variety of service providers. These services all operate simultaneously over the same DSL line without getting in each other's way, and some have requirements for quality of service. The reports define the complete end-to-end architecture for this network, and specify the operation of each of the key elements -- eventually working down to a level of detail permitting formal testing.

What comes across loud and clear is the explicit goal of developing technology to compete effectively with cable. This is mainly a North American requirement -- outside North America, DSL is far ahead of cable in the battle for broadband subscribers. The telcos have already deployed a lot of expensive technology; the challenge is to put new technology in place while maintaining as much as possible of the existing underlying technology.

We'll describe the new strategy by reference to the reports, which are all public. The current versions of these reports, and many more, are available for download from .


Overall Architecture

The architecture is described in two reports, TR-058 and TR-059, both published in September 2003. Taken together, these establish the framework for the next generation of DSL services designed to meet the needs of network and application service providers offering a wide range of services to customers.

TR-058: Multi-Service Architecture & Framework Requirements (http://www.dslforum.org/aboutdsl/Technical_Reports/TR-058.doc) describes the overall architecture, the business rationale and the roles of the various players: Network Service Provider, Application Service Provider, Loop Provider and Access Network Provider. The key to the architecture is the provision of "many-to-many access through a common Regional/Access network" shown in the following diagram. A variety of services from many NSPs and ASPs pass simultaneously through a common interface to several users with different devices in the home.


Figure 11 - Many-to-Many Access (from TR-058)

TR-059: DSL Evolution - Architecture Requirements for the Support of QoS-Enabled IP Services (http://www.dslforum.org/aboutdsl/Technical_Reports/TR-059.doc) moves down a level and describes the reference architecture, including how QoS is provided for those applications which require it. It describes the need for several different kinds of QoS, ranging from a "turbo" button for a customer wanting to speed up downloading, differentiated services to provide priority for business customers paying more for DSL, to real-time applications like VoIP and streaming video which require tight control over jitter and packet loss. Since DSL is based on ATM and most applications today are based on IP, it defines how the IP QoS mechanisms are handled by the underlying ATM layers. Although the ATM infrastructure is "non-IP aware", it describes how traffic engineering of the ATM layer can support "IP-aware" functions in the network and the customer premises.


The BRAS, the DSL Gateway and the Home Network

Two devices play a central role in the new architecture: the "Broadband Remote Access Server (BRAS)" at the core of the network, and a "DSL gateway" at the customer premises. The BRAS is the connection point to the network and application service providers, while the DSL gateway is the connection point to the network in the customer premises. These two devices work together to implement the architecture of TR-058 and TR-059 while preserving the existing deployed DSL infrastructure.

The BRAS is defined in TR-092: Broadband Remote Access Server (BRAS) Requirements Document (http://www.dslforum.org/aboutdsl/Technical_Reports/TR-092.doc). Published a year after TR-059 in August 2004, TR-092 goes somewhat beyond TR-059 reflecting additional thinking about the supported services and their requirements. It envisions an elaborate regional server for large central office deployments: "A large CO device will typically support between 64K to 128k subscriber sessions and an aggregate downstream bandwidth of 2.5 Gbps."

Its peer on the customer premises is the DSL gateway described in TR-068: Dual Port ADSL Router (DSL Gateway) (http://www.dslforum.org/aboutdsl/Technical_Reports/TR-068.doc) published in May 2004. This document completely defines a residential gateway device down to the location, labeling and color of the front panel lights and the color of the ports on the rear.

Finally, the most recent report TR-094: Multi-Service Delivery Framework for Home Networks (http://www.dslforum.org/aboutdsl/Technical_Reports/TR-094.doc) describes how home networking completes the picture. It defines "a functional home networking architecture that permits multiple residents within the home to use multiple applications and devices with differing connectivity requirements (QoS) and at the same time minimizing poor application performance that could result from conflicting or competing application demands."


Next Steps

These reports, and others on the DSL Forum site, go a long way to defining the new architecture. We understand more reports are in process to further flesh out the new architecture.

Since the Digital Living Network Alliance (DLNA - formerly DHWG) has been working on many of the issues in TR-094, we asked Michael about the relationship between DSL Forum and DLNA. He told us that the service-provider members of DSL Forum have played a major role in defining these requirements, and that consumer electronics companies are now "recognizing where the service providers are going." While DSL Forum is not itself a DLNA member, he said "a group of DSL Forum members are working with DLNA."

Michael pointed out that the DSL Forum architecture "is useful not only for DSL - the same specs can be applied to any other access mechanism." As telcos roll out new flavors of DSL -- ADSL2, VDSL, VDSL2 -- and several variations of fiber including fiber to the home (FTTH), most elements of the new architecture can be applied.

Now that these documents have been published, the next step is for the telcos to write RFPs based on them and for vendors to respond. We have been told that the telephone companies participating in the DSL Forum are moving forward to implement this new approach; we note active involvement from Bell Canada, BellSouth, BT, SBC and Verizon. There may well be some changes as the telcos and their vendors iterate the requirements and the associated costs.

The new architecture suggests that the telcos' approach to the "triple play" may be very different from that of the cable operators. While it is not called "open access" it certainly comes close -- allowing many providers to offer whatever services they want to provide, over a common network, with whatever specific facilities they need.

A big unknown in this new approach is whether network and service providers really want to deliver the types of services contemplated in the new architecture -- and whether they are willing to pay for the network delivery services, especially QoS. There have been many discussions of the "open access" model, and many companies have been pushing for it for years. Now they should be willing to come forward and participate.

While the implementation of this new approach will take some time, and there will undoubtedly be some changes along the way, DSL Forum certainly appears to be providing its members with the framework for DSL to become a formidable competitor to cable. At the end of the day, results are what counts. We’ll all need to keep watching as the competitors contend for customers and their money.

( www.dslforum.org ) ( www.verizon.com ) ( www.bellsouth.com )