Columbus, Ohio.... ....and Boulder, Colo.

Sometimes, technologies which work similarly under the hood are put to use for different purposes. Those different purposes are driven by the different strategic motivations of their inventors. Satellites, for example, all orbit the Earth in some way, but are used for communications and entertainment; ditto for the humble radio itself.

Such is the case with the technologies used by AT&T Broadband and Time Warner Cable to connect non-affiliated Internet service providers (ISPs) to their HFC plant. The major parts–back office interfaces for provisioning/registration/customer service, policy-based routers–were built in a way that's probably similar enough for the wider industry to adopt. That, in turn, contributes attractively to economies of scale.

There are nuances, however, mostly related to different underlying mindsets of both providers–and their corporate parents. Time Warner's new daddy is an ISP, and the largest one to boot. AT&T Broadband's relatively new daddy is a telco–the largest such beast. Understanding the root motivations of the biggest ISP and the biggest phone company illuminates the reasons for the many nuances that differentiate the open access tests in Columbus, Ohio (Time Warner), and Boulder, Colo. (AT&T Broadband).

Motivational drivers

Columbus skyline photo courtesy of the 
Columbus Convention 
& Visitors Bureau. 
Photo by Rod Berry.

For Time Warner Cable, it's all about getting at least one outside ISP onto its broadband plant quickly and smoothly. It isn't optional. Under the terms of the Consent Decree that accompanied the AOL/Time Warner merger, AOL isn't allowed on Time Warner's plant until some unaffiliated ISP hops aboard first. That somebody else (in Time Warner's case, Earthlink) has to come aboard via an established commercial relationship. That means money has to change hands via a service level agreement (SLA). The sooner Earthlink gets on, the sooner AOL gets on–with interesting possible effects on cost.

By contrast, AT&T's motivation is to use open access, which it brands "Broadband Choice," as a way to attract as many customers as possible to its high-speed Internet service. Whether those customers come in at wholesale (via outside ISPs) or retail (home-grown, whether cable modems are leased or purchased by consumers), the answer is "yes." A high-speed Internet customer, to AT&T (and other MSOs) is a two-way Internet Protocol (IP) customer. A two-way IP customer is a prime target for other advanced services, like voice, and in time, interactive video.

AT&T Broadband isn't bound by a Consent Decree, but its motivations are just as constitutional. With the equity angst of its corporate parent, and the intent to take the Broadband division public sometime this year, the need to attract more advanced service subs is acute.

The similarities

The equipment that's the most costly, and therefore the better candidate for common use among MSOs (to drive prices down), is the policy-based router. Both AT&T Broadband and Time Warner Cable use one. The "policy," again in both cases, is to examine the source of each user's packets in order to route them to the proper ISP.

Back office interfaces, for provisioning/hookup, customer service, trouble-ticketing, etc., may be different now, but ultimately, they'll be the same. It's plausible that CableLabs will initiate coverage of open access, focusing especially on the establishment of common, back-office interfaces. Technological unity on the big-ticket items and the interfaces is good news for MSOs who haven't yet dabbled in open access, but may do so later.

The differences

There are three areas where AT&T Broadband and Time Warner Cable differ in their approach to open access. The biggest relates to how consumers attach and detach themselves from the HFC plant. Another is in the demarcation between ISPs and the cable plant. A third is how IP addresses are managed.

1. Connecting/disconnecting. The bulk of AT&T's work centers on a piece of software, called a "service agent," that is installed on a consumer's PC when he connects. Its purpose is to make it easy for consumers to switch ISPs or resolve troubles. A mouse click connects the consumer, via the Web, to an ISP's registration screen, where each ISP's screen has a similar look and feel.

Flatiron Mountains photo courtesy of Boulder Convention and 
Visitors Bureau.

The service agent was home-grown by AT&T Broadband, and involved the development of interfaces that can be hitched to the billing, customer service and other back-office elements of participating ISPs. Want to connect? Use the service agent. Want to switch? Service agent. Problems? Service agent.

On the surface, AT&T's service agent may appear risky: An invitation for consumers to churn. But, recall that wholesale economics improve with volume. The rationale? Even if consumers switch to another ISP, some portion of their monthly connect fee still goes to the wholesale provider (AT&T). In that sense, the more ISPs that hop aboard, the better. And the more customers they bring, the better.

Plus, the service agent includes a notable consumer-retention tool: A sort of centralized e-mail "bucket" that lets subscribers go to one place (AT&T Broadband) to access all their e-mail accounts. Most Internet users have at least two: one for work-related communiques, and another for personal use. Many are loath to do anything that requires changing their e-mail address, because it's a hassle. But if all e-mail accounts are a click away, in a central location–that's added value, and a useful retention tool.

Time Warner, in part because of the Consent Decree, thinks differently about consumer connects/disconnects. For starters, there is no service agent. Consumers who sign on do so by phoning or visiting the Web-based registration page of the participating ISP. Disconnecting is parallel to traditional churn, and isn't made easy. If an Earthlink/Time Warner customer wants to switch to, say, Juno/Time Warner, it means calling (or e-mailing) Earthlink, disconnecting, then connecting to Juno.

2. Lines of demarcation. Time Warner, again in part because of the Consent Decree, identifies the demarcation point as the headend. AT&T doesn't use the term "demarc," and instead accepts outside ISP traffic at "points of connection," either logical or physical. Specifically, AT&T accepts traffic from participating ISPs in three ways:

Over the ISP's own backbone resource, to a mutually agreeable point-of-presence; over Excite@Home's backbone, at a regional data center connected to a mutually agreeable POP; or over a mutually agreeable POP on AT&T's backbone–AT&T Worldnet runs POPs just about everywhere–which links to an Excite@ Home regional data center.

These differences could affect cost, but not necessarily technology. AT&T hasn't yet formalized commercial arrangements, but appears to want to charge ISPs some fee for data backbone access.

But what if Earthlink, whom AOL is dependent upon to gain access to Time Warner networks, determines that backbone costs are going to be too steep for its liking, and retreats from the project (or threatens to)? And what if Time Warner, under time pressure from AOL, gets too "creative" with backbone costs–which is to say, offers a backbone ride at its own cost?

Backbone costs are the single largest cost component of both Road Runner and Excite@Home. They are the reason ISP Channel shut its doors. Agreeing to carry outside ISP data at MSO cost would do nothing to ease backbone costs. Done too soon, it would also set a low financial bar for any other MSO that wants to etch a similar Earthlink deal.

3. ISP address blocks. A minor difference between AOL TW and AT&T relates to IP address management. Time Warner's pattern is to manage the IP addresses that are delivered from the external ISP. AT&T will do that, or assign IP addresses from its own private reserve. Again, this isn't a biggie for the industry-at-large, but an indicator of different corporate philosophies.

Trial status

In Boulder, AT&T is wrapping up a test that included about 300 customers within a footprint of about 18,000 cable modem homes. Ten ISPs were invited; eight came in over the transom. Just four were able to make it work: AT&T Worldnet, Earthlink, Juno and Excite@Home. Testers were asked mostly to flex the service agent by switching ISPs often. Initiated in October 2000, the tests end this month. A parallel effort is underway with AT&T and participating ISPs to move into commercial rollout. A follow-on launch happens in Massachusetts by Halloween, followed by a statewide Broadband Choice launch by next summer.

In Columbus, "fewer" customers, mostly "friendly homes," are participating in the tests, according to company officials. Internally, the project is viewed as one of the most important corporate priorities, and may even surpass video-on-demand as a 2001 launch priority. Additional markets haven't been determined yet, but it's plausible to expect at least three more systems will be opened by year-end, if not more.

What matters now, in open access, are the business arrangements that emerge, through service level agreements with participating ISPs. Time Warner, because of the consent decree terms, is the one to watch. When you're under the gun from corporate to get something done fast, you're more approachable for "creative" deals. But, what's creative for Time Warner may not be appetizing to other MSOs.

Other MSOs need to focus on open access from a strategic level. That means examining the economics of wholesale and retail traffic. It also means knowing how to track traffic, so as to charge ISPs by speed requested and total bit consumption.

It's likely that the business models for open access, not unlike the business models for satellites, radios and other historically significant technologies, will be quite different among the service providers. AT&T likes wholesaling because of its effect on total network usage. AOL, through Time Warner, sees it as a way to push more of its media through a wider, tightly controlled pipe. Others may prefer some hybrid.

One thing is certain: It's a lot more complicated than in the days (like now) of one ISP. But the underlying technologies are similar enough to be non-alarming.

Author Information
About the author
Leslie Ellis is an independent cable technology analyst and author based in Denver, Colo.