Netwatcher (ISSN 0890-5800) is a monthly publication of CIMI Corporation. Subscription information is available here. Copyright © 2000, CIMI Corporation. All rights reserved. No publication or reproduction of this document is permitted without the express written consent of CIMI Corporation.

Management Briefing

This section is being omitted this month to provide space for the CoSine  VPN response offered to our subscribers.  Sorry!

In the Know

Last fall, the FCC took a revolutionary step in its attempt to promote the goals of the Telecom Act.  Recognizing its legal mandate to promote the congressional objective of universal residential empowerment with broadband access, the FCC made a simple bargain with the incumbent LECs, particularly the RBOCs.  “Deploy broadband access based on frame/cell technology,” they said, “and you’ll be shielded from unbundling whatever’s inside the cloud”.

That order goes into effect in May—the month you’ll first read this issue.  Despite this, we’d probably be sitting around reading tea leaves to gain an understanding of the RBOCs’ deployment plans if it weren’t for a happy accident.  SBC was involved in a merger request (with Ameritech, obviously), and as a result of this merger was required to file documents with the FCC on the nature of their new network.  In short, SBC has announced how they’ll respond to the FCC’s policy.

Perhaps “announced” isn’t the right word, though.  A regulatory filing is certainly a public announcement in one sense, but it’s not exactly like standing on the Ball in Times Square and yelling through a bullhorn.  While SBC did make a public announcement of its overall strategy in its “Project Pronto” material (see their web site), that material doesn’t go as far as their FCC disclosures do.  The press, with a few minor exceptions, hasn’t covered the FCC material in a comprehensive way.  So…we’re going to cover it here.  In addition, we’re going to project from that material what is likely to be the long-term direction of the RBOC networks.

As our readers know already, the gist of the FCC policy statement of the fall was that RBOCs could shield packet (meaning frame/cell versus TDM) plant from unbundling, but they’d still have to wholesale the services of the network and the customer loop.  This combination has led to the RBOC decision to deploy an ATM-based network with fiber remotes very close to the user.  Such a network devalues the wholesale loop by shortening it and reducing the concentration of customers at any point of access (from ten thousand or more in a CO to about 700 in a remote).  It devalues wholesale service by making sure that the “service” of the network doesn’t have a large retail market, which ATM doesn’t.

SBC’s plan, as filed with the FCC, involves the use of Alcatel Litespan 2000, a fiber remote that uses OC-3 uplinks in its current version.  The remote end of the 2000 will go in a hut or vault in place of a standard fiber remote (TR-0008 or GR-303), and the central office terminal (COT) side will be connected to an ATM switch (Lucent’s, presumably, but no FCC disclosure in this area was required so we don’t know for sure).  This switch will link each of the COTs with the Class 5 Anymedia switch by Lucent (SBC did a big deal on those in 1998) or other Class 5s, and will also provide CO-to-CO and virtual tandem services.

In each Alcatel 2000, copper loops will be terminated and a splitter will separate the digital and analog spectrums, as required in the FCC “line-sharing” ruling (99-255, also last fall).  The analog portion will be pushed through a PCM encoder and cellularizer, emerging as an ATM cell stream.

An OC-3 is reserved to carry the baseband voice up to the CO.  This OC-3 is divided into what is effectively three T3s, serving a total of 2,016 voice lines.  This permits the OC-3 to carry all the voice cells without congestion.

For the digital bandwidth of the loop, SBC is supporting ADSL initially, and is promising the full 6 Mbps downstream bandwidth.  The company estimates this can be delivered only with copper loops of about 3.5 miles or less, and this gives them an excuse to trash a lot of home-run copper loops, replacing them with fiber remotes that offer better wholesale protection.

The digital portion of the loop is routed as a single ATM PVC, and all the DSL-PVCs are routed onto a second OC-3.  In effect, this oversubscribes the “data” OC-3 by about 10:1 in the customer-to-net direction and by almost 100:1 in the downstream direction, assuming every customer gets DSL.  Based on our estimates of DSL penetration by loop (not by customer), the actual oversubscription is between 8:1 and 20:1 at worst.

All this talk about oversubscription is important because SBC’s filings indicate it’s offering only ATM UBR PVCs in its early offering.  By routing the digital portion of the loop to an oversubscribed OC-3, SBC is insuring that cheap data-directed bandwidth isn’t arbitraged for voice service, undermining its voice profits.  While that may seem a cynical step, it’s essential if the broadband deployment is to proceed.  Nobody is going to invest in capital projects to lose money, after all.

Two thousand voice lines seems like a lot, but it won’t be adequate for higher-density sites, particularly when business voice services come on line.  For those services, SBC is proposing to use SDSL, a symmetrical DSL that offers about a megabit of bandwidth in each direction.  This is enough for 16 voice lines using standard 64 kbps modulation, and SBC has indicated that SDSL will be their strategy of choice for multi-line business voice delivery except in the largest sites.

SBC needs to get Alcatel’s SDSL deliverable on the Litespan box, of course, and it’s likely that this will be accompanied by an improvement in the uplink speed.  The RBOCs in general are said to be looking for an OC-12 and OC-48 option on their remote gateways, both to allow for higher customer density and to support QoS-specific business services at a later date.

The issue of higher QoS is a very delicate one for SBC and the RBOCs.  The problem is that a high-QoS link would present a risk to their voice-handling revenues, but it’s clearly a necessity if the RBOCs are to support new services.  The standard Internet service quality is sufficient for only about a $30 billion market; the rest depends on better quality.  The key for the RBOCs is striking a balance between new revenue gain and voice revenue loss.

The only way to achieve that balance is through the timing of the formation of their Advanced Services Subsidiary (which we’ve abbreviated “Asub” to avoid the obvious acronym).  Asub is going to be the vehicle for the RBOCs to enter the data market without incurring a wholesale risk.  It will acquire virtual circuits from the ILEC and add data infrastructure to create high-margin services, so it’s dependent (like the CLECs who will take their marching orders from its relationship with its parent) on the nature of the access services available.

The balancing act the RBOCs perform is in the timing of Asub in relation to their access modernization.  Since Asub won’t have any real revenue sources until it can sell value-added services based on the ILEC component’s deployment of broadband access, it’s important to create Asub when the ILEC has some control over the pace of access reform.  That probably means late this year or even early in 2001, since the Alcatel modernization of the Litespan won’t be available until this fall.

With faster uplinks, the fiber remotes could begin to absorb not only analog connection upgrades to a form of DSL, but also displacement of existing private lines.  The RBOC immunity from unbundling the elements of their packet infrastructure doesn’t protect fiber that’s deployed in TDM missions (in fact, it couldn’t be applied to any purely business service elements, only business services offered out of an infrastructure used for residential access as well).

Private line displacement is a knotty issue for the RBOCs, too.  On the one hand, most T1, T3, and SONET deployed by ILECs for customer use are either used by businesses to access an IXC’s POP (which benefits the rival IXCs),  or by a value-added carrier who will ultimately compete with the Asubs.  Thus, it would make good business sense to kill the TDM stuff off; it might even cause some customer churn that the RBOCs could benefit from.  On the other hand, killing business service access will clearly aggravate a lot of big-spending firms with good lobbyist support, and it might backlash against Asub as well, or even spawn a CLEC opportunity to provide TDM that the RBOC can’t or won’t provide.  That could force the RBOCs to discount their ATM-based access, lowering margins and increasing the leverage a wholesale player would get from obtaining a high-bandwidth business connection.

For now, since T1 and T3 fulfilled over repeated copper has little wholesale risk (it can’t be leveraged) and since SONET access is rare and expensive enough to reduce its wholesale benefit, the ILECs may stay the course.  In the long term, they’ll surely convert everyone to ATM access.

That brings up the timeline of all of this.  Here, we’re completely in the realm of speculation, but some of the issues seem clear enough to justify going a little out on a limb.

The first step the ILECs will take, likely this summer, is the announcement that all future provisioning of residential services and voice services to smaller business sites will occur only through their ATM outside plant.  This eliminates the risk that a CLEC will demand fiber be provisioned under the terms the RBOCs currently provision it to TR-0008 and GR-303 remotes.  In effect, it kills off “fiber on demand” for the CLECs.

This first step only generates new broadband service for those businesses and residences who require a new remote, which isn’t a large number and which can’t be easily controlled in terms of pace of growth.  Thus, it won’t fuel the revenue needs of Asub.  To do that, the RBOCs will have to begin conversion of existing remotes to the new ATM-based remotes.  This isn’t likely to happen until Alcatel can supply the new gear, which we expect won’t be until about October.  Our view is that sometime between then and year-end, the second phase of remote replacement will begin.

The third phase, likely to begin about mid-2001, is the replacement of selected home-run copper with fiber remotes.  Up to now, the impact of this on DLECs would have been indirect (the competition would drive DSL prices down, reducing their margins), but pulling the copper out of service will reduce the number of loops that a DLEC could hope to convert to DSL.  We think that the DLEC market will have long-since transitioned to a service play, but at this point there’s little choice.

At about the same point, we think that the RBOCs will begin to open their ATM services beyond the UBR level, leveraging the revenue gains of Asub to offset incremental risk of loss of voice revenues.  This will also mean the beginning of integration of the current business access connections (T1 and T3) into the new ATM network.  We expect that whole process will take about a year, making the full transition of business something we can expect in the second half of 2002.

The pace of DSL rollout to residences and smaller businesses will probably require longer.  Our assessment is that the total DSL prospect base, absent Universal Service subsidization, is about 60 million lines.  The proactive target portion of this, including small business and telecommuters, is about 44 million lines.  We would expect to see about half of these deployed by 2005.  By 2008, we’d expect to see DSL penetration approaching the 60 million level, and with some subsidization likely by then, total DSL penetration in the US to reach something like 70 million lines after 2010.

During this period, the business model of the RBOC will become more and more polarized, with Asub first doing the data service overlay stuff, then (by around 2004) all retail service sales.  By mid-decade, we expect to see Asub spun off as a separate company to protect it from having to make retail profit contributions to the profitability of the increasingly-wholsale-dependent ILEC.

This discussion is primarily centered on SBC's plans, because they're visible.  We expect that Bell Atlantic will follow this model closely, and BellSouth will probably differ more in vendor selection and possible greater reliance on ATM PON structures for its remotes.  US West, tied up in a ménage a merger, is too up in the air to call at this time.  We'll be watching all the players as they bring their strategies forward, and we'll update this material as required.

Strategies

CoSine has delivered a VPN response that combines virtual routers with deterministic core network support, so it's quite interesting.  Sorry, but it's for subscribers only!

Down the Line

Next month, we’ll look at the economic issues facing carrier competition and new infrastructure roll-outs.  We’ll continue our VPN vendor submissions a bit longer this spring, then publish the entire group and our three-part “requirements” piece as a PDF.