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

Optics is hot, but the hottest topic in optics isn’t even much recognized by the media.  It’s passive optical networking, or PON.  There’s a good chance that most businesses will be touched by PON in the next five years, and almost a unity probability that PON will touch every business by the end of the decade.

PON is a fiber architecture in transition, its original application having been largely displaced by newer ones.  Still, both the PON of old and the PON of today are alive and well, and there’s a good chance that PON will eventually influence not only business communications, but residential voice and data, and even entertainment video.

The original PON objectives were more related to broadcast television than to two-way communications.  Starting in the late 1980s, various trials were conducted on the use of fiber architectures that served multiple consumers using passive splice technology (hence the name) rather than a network of active devices.  This would cut the cost of fiber deployment sufficiently to permit its use as an alternative to CATV cable in provisioning television services.

In application, PON creates what might be called a “tree”, a tributary structure that involves splice-based “forks” that eventually terminate in consumer locations.  The tree is fed at the trunk or head end with enough laser power to not only accommodate the optical dispersal along multiple paths, but to account for the loss in the splices.

A PON tree can be used to carry virtually any signal that can be converted into optical laser drive, including analog and digital video.  However, the nature of PON is that the tree works fine from trunk to branch and less well from branch to trunk.  The reason is that traffic originating at the branches would collide at each splice point, contention.

Solutions to the PON contention problem have been proposed for some time, but only one has really been broadly accepted.  APON, or ATM PON, is a cell-based architecture that employs TDM multiplexing that some will find reminiscent of the old SMDS slot structure.  The architecture allows the optical tributaries on the tree to receive a common feed (the downstream direction) and to share an upstream feed to the head end.

APON has been standardized (G.983.1), and large-scale trials are already underway.  APON is also apparently the architecture that most of the RBOCs will deploy to provide for broadband access to business and residential services.  It’s suitable as a direct business service (ATM is then the payload to the user) and as a feeder for both ADSL and the higher-speed (up to 58 Mbps) VDSL.

The G.983.1 standard for APON is a fairly conservative one.  The architecture consists of a tree with a total of 32 splits (64 endpoints) and a total length of no more than 20 km (12 miles).  One mode supports symmetrical 155 Mbps (OC-3) service, and another has a 622 Mbps downstream feed (OC-12) and a 155 Mbps upstream feed.  The electrical layer supports SVC capability as well as PVCs.

A more aggressive approach is being taken in the so-called “SuperPON” trial, in which Alcatel is the main player.  SuperPON increases the service data rate to 2.5 Gbps (OC-48) downstream and 311 Mbps (OC-6, if you can imagine) upstream.  The topology of SuperPON isn’t entirely passive; there’s a feeder and feeder repeater structure feeding amplified splitters.  The feeder-to-splitter distance can be as much as 90 km (54 miles), and the terminations can be up to 10 km (6 miles) from the splitter (the 90/10 topology is also supported for APON, but not by everyone).

Those interested in how APON and SuperPON work at the electrical and physical layer can refer to the relevant standard and other documents from the vendors for the details.  At the high level, the downstream feed is TDM-divided and its upstream feed is controlled by TDMA (time division multiple access).  Broadcasting downstream is easy, and redundancy can be supported through the use of doubled fiber strands.  Studies generally show that PON is more fault-tolerant than SONET, because the components in PON aren’t active and can’t lose power.  The MTBF of a splice is also clearly lower than that of a node.

The international Full Service Access Network Consortium (FSAC) has declared APON to be the most effective approach to generalized broadband access services, which it clearly is.  That it’s also a player in the new-gen RBOC infrastructure has been noted already, though the exact role and timing of PON there may be still undecided.

For the RBOCs, PON may solve a knotty problem; how to get fiber infrastructure into the field at a large scale without an unacceptable wholesale risk.  ATM infrastructure is shielded from unbundling, our readers will know, but customer loop isn’t.  Here’s a critical question, then; is PON loop shielded?  Even if the answer is that it is not, what’s the wholesale price and what are the restrictions on access?  You can’t co-locate with an optical splice, after all.

The RBOCs appear to have decided that at the very least, APON offers no more risk of unbundling than other fiber remote strategies, and SBC has specifically indicated that its Project Pronto infrastructure will serve large businesses (and presumably multi-tenant sites) with APON.  Less clear is the nature and timing of the use of PON in the residential market.  The newer Alcatel Litespan 2012s have PON capability, and thus could support the deployment of APON-based residential services.

If SBC intends to shield their business infrastructure from unbundling, as we’ve suggested, it’s our view that ATM (with or without PON) isn’t enough.  There’s no statutory basis for claiming exemption for pure business infrastructure, since the FCC ruling involved (99238) draws its authority from the Telecom Act’s mandate of the FCC to promote universal service—meaning residential service.  Thus, SBC must associate residential service with any given PON tree to claim exemption from unbundling.

But this means that any business PON tree must have residential subscribers, so some neighborhood gateway remotes must be based on APON.  Will they all do so?

We think most will, perhaps all.  The problem for SBC and the other RBOCs is that one never knows when and where a business user will pop up.  Deploy specialized residential fiber remotes with no PON capability and you risk deploying something you can’t leverage without rebuilding it.  PON also lets SBC and others create more use for access fiber that might be lying around in dark fiber form.  The 10Q Bell Atlantic filed with the SEC this spring notes that the same FCC order that creates the unbundling exemption also mandates dark fiber be unbundled.  Why do that unless you plan to light it?

In any event, it should be clear to all that PON in general, and the APON/SuperPON family in particular, are things to be watched for the future.

In the Know

What will save the CLECs?  That’s a question that has been raised from time to time since the term “CLEC” was spawned by the Telecom Act in 1996.  Today, as CLEC stocks seem to be slipping, there may be a new urgency to the survival debate.

The issue is complicated by the fact that the media/marketplace has adopted the term “CLEC” to delineate a whole class of players that aren’t even conformant to the Act’s definition.  Add the fact that there is strong evidence that “CLEC-hood” is more a role than an exclusive business model, and you’ve got enough confusion to obscure an army of truths—which, of course, we don’t have.

Hopefully, we’ll add some truths here, but we can’t be sure that all the angles are visible yet.  The market is still shifting, folks.

What’s A CLEC Anyway?

According to the Telecom Act, a Competitive Local Exchange Carrier is a carrier who competes with an incumbent LEC in providing local telephone services.  It’s clear from the context of the Act that the FCC and Congress expected CLECs to offer local voice calling, as the majority of the wholesale and interconnect issues covered in the Act relate to that type of service.

Did they expect a CLEC to be exclusively a CLEC?  The wording of the Act doesn’t make this clear, but it’s hard to see how regulators and legislators would have thought that a competitor could achieve profitability in the local exchange space when incumbents were earning less than 15% on investment and the public expected widespread cost reductions.  Thus, we believe that “CLEC” defines a role a broader-based carrier would play, in addition to its primary revenue-generating role.  In fact, it could be argued that the FCC and Congress thought that the CLEC role would be played primarily by the current IXCs, who would jump into local service to counter the ILECs’ release into the long-distance market.

Truth and logic aside, the market definition for the term “CLEC” has broadened in stages.  First, it was applied to DSL-exploiting competitors who acquired wholesale copper loop assets and co-location space—firms who now tend to be called “DLECs”, with the “D” standing for “DSL” or “data”, depending on who’s talking.  Finally, it’s been applied to the fiber-based competitors who primarily target multi-tenant and large-company facilities.  This group is the “BLEC”, where the “B” is “broadband”.  In short, we’ve made anyone who provides access into at least a member of a CLEC subclass.

Grizzled veterans of the market will recognize that the so-called “BLECs” are really what were once known as “competitive access providers” or CAPs, since their business strategy, technology, and target markets are virtually identical.  One could also argue that the DLECs are also CAPs, since they are also focusing on selling access to the services provided by others.  At the very least, all of these players seem to have a CAP value proposition somewhere in their core.  CAPs, after all, provide alternative access, and that’s what all this stuff really is when you strip off the covers.

Trying to push the CLEC classes into the CAP category isn’t just an exercise in semantics, though.  We have a decade of experience with the CAP business model, and this experience can serve us as a framework for judging how CLECs will have to position themselves to survive.  While the new regulations created by the Telecom Act and subsequent FCC orders changes the market picture, it doesn’t alter the fundamentals of the business process.  We can adjust for the changes these forces bring; they won’t be completely revolutionary.

Access Revenues and You

An access network provides a customer set with a means of getting information between customer sites and what might be called service points of presence (SPOPs).  A leased-line connection from a site to an interexchange carrier POP would be an access line, in short, and that’s exactly what it’s called.  Thus, the term “access” has come to mean a customer connection into a service network, not the service network itself.  It follows that the access connection and network are simply conduits for traffic.

Access competition has historically been focused on price differentiation, because service access tends to be featureless (making feature differentiation rather challenging).  CAPs have thus been forced to consider price/cost relationships very carefully to keep their margins up.  A result of this focus is the tendency for CAPs to operate only in high-opportunity-density areas like major cities.  There, a mile of fiber passes a large number of prospects.  Network costs don’t rise proportionally to customer count because of this, yet revenues do.  Thus, the business gets more return on capital as it gets more customers.

Despite this, we haven’t seen huge CAP successes.  Teleports, fiber access CAPs like MFS, and newer regional access fiber players have all been uninspiring in terms of revenue return.  To be sure, they’ve cited the “explosive growth of the Internet” and the “corporate demand for infinite bandwidth” as indications that their future is brighter.  Yeah, maybe—but those factors have been cited for a decade and they haven’t come through yet.

In fact, we have some real reasons to think that the trends might be moving in the opposite direction.  Until now, access competition has focused on large consumers.  With the advent of specific regulatory pressure to secure universal broadband access, the requirement for infrastructure breadth develops.  CAPs, you’ll recall, were founded at least in part on the assumption that the best strategy was infrastructure focus.

In fact, we could say that the DLEC of today is the CAP of yesterday, in an inheritance sense, even though the BLEC looks more like a classical CAP.  The DLEC is attempting to reflect in its business model the regulatory imperative of breadth of market.

That’s where the problem comes in, of course.  The classical T1 line of old sold for about a thousand bucks a month, and competitive pressure has brought the average price below $500 in most areas.  The classical DSL service is less than half that, even for businesses, and yet the service is marketed to a much broader customer base—meaning the infrastructure to offer the service has to cover more territory.

So what’s the profit model for the new xLEC?  A number of options could be explored:

1.       The CAP model of focused infrastructure and a high-price-tolerant target market.

2.       The DLEC model of broad-based demand, and broad-scale infrastructure for supplying it.

3.       The “Access-plus” model where higher-margin services with differentiable features add to the base-line access revenues.

Some may be concerned that we seem to be throwing out the popular “wholesale” and “retail” CLEC models.  Well, maybe we are, so we’ll give a quick defense of that stand before we move on.

A “wholesale” carrier is someone who sells not to the end consumer but to another carrier, where a “retail” carrier sells to the consumer directly.  Since all carrier dollars ultimately originate with the consumer/business, the only difference in these two classes is how they manage their costs.  Wholesale players focus on infrastructure and skimp on sales/marketing costs, in return for paying what amounts to a sales commission to get customers.  Retail carriers have direct customer relationships, affording them potentially higher profits if they build out their infrastructure, but exposing them to higher infrastructure cost.

In the modern market, the real differentiator is a little different.  A retail provider must contend with the issue of developing access demand.  In general, only sophisticated buyers would be able to integrate access connections to real business services.  Try to sell residential DSL without any Internet or other service to pull it through!  Thus, a retail access player is almost forced into being an “access-plus” player.

A wholesale access player, in contrast, simply partners with a retail carrier that offers an attractive service (Internet access, for example), letting that service pull the access service through.

OK, then.  We might well say that a wholesale and retail xLEC differ from one another only in whether the xLEC owns the magnet service that attracts the buyer, or obtains it.  In effect, both are “access-plus” players.  Given that the wholesale/retail issue dominates the market today, does that say something about the CAP and DLEC models of profit?

You bet it does.  We believe that the CAP and DLEC models are probably not going to return enough on capital investment to secure a survivable market position for their proponents.  The problem is that the focus requirements of the CAP model are not easily defended in a marketplace where public policy demands universal broadband, and the breadth of the new access market makes it difficult or impossible to control cost of infrastructure.

Does that mean that all these so-called CLECs are dead?  Maybe not, but they’re infected.  Their survival depends on a somewhat personalized cure strategy.

Cure Strategies

We can classify xLEC strategies into two groups; the flip-seeking and the business-seeking.

Flip-seeking xLECs must attempt to build market value quickly without creating the impression that they’re in a niche whose long-term value is minimal.  Most flip-seekers have attempted to value themselves on the collective value of their customers.  This is particularly true in the DSL/DLEC space, because the theory goes that these customers will be exploitable in the future and thus have economic value to a prospective buyer.  Valuations of as much as $15,000 per residential customer have been set.

For the flip-seeker, the key would then be to create customer impact as quickly as possible, despite the risk of building too much infrastructure and creating a bad rate of return.  This is because “rate of return” won’t be too much of an issue in a market that values the company on a per-customer basis.

Given the nature of this space, we believe that flip-seekers should be delighted at the prospect of being able to wholesale xDSL access via ATM remotes from a new-age RBOC infrastructure.  This would generate customers by eliminating current constraints on co-location, copper loop quality, etc.  If customer valuation remains the metric of judgment on xLEC survival and value, wholesale DSL is a drop-dead winner.  Applied quickly, it could build value without increasing expenses tremendously, but at the expense of return on capital.

Not all xLECs in the DSL space will see it our way, though.  Some have VC investors that are also linked to DSL equipment vendors and won’t surrender the hope that CLECs will buy hosts of DSLAMs to bail out their position in the DSL startups.  Others may simply think the RBOCs won’t really deploy DSL in as aggressive a way as they promise.  We don’t think those are survivable views.  You cut your capital expenses through wholesaling, build customer base by extending the marketing programs, or perish.

The same strategy is probably relevant to the BLEC.  Getting glass into a lot of buildings quickly creates the impression of a land-rush business model (they aren’t making any more land, and they aren’t making too many more multi-tenant buildings, either).  There is some indication that the per-location-valuation metric will prevail in this market, as the per-customer-valuation metric has in the DSL space.

Business-seeking xLECs have to be focused on return on capital, which means focusing on climbing the service value chain.  The problem is that it’s not clear how to do that.  IP VPNs are about the only widely accepted new-gen data service, and they have a poor near-term revenue realization rate because of market ignorance and lack of carrier direction.  Clever work could turn this around, but not without difficulty.  The successful data services are primarily ones that have little or no local or regional credibility—frame relay, for example.  There are regional customers for these more traditional data services, of course.  They’re customers of the RBOCs, in the main.  Thus, taking on the incumbent LECs is virtually essential in building a long-term position.

Which approach is best?  The truth is that neither will probably save most of the current xLEC players, because none have structured on either of these models, most won’t recognize the need in time, and most don’t have the capital reserves.

Bad things are going to happen in the “CLEC” space over the next year.

Strategies

 Spring Tide Networks is the last vendor to submit a VPN response in our ongoing review of VPN products.  Perhaps it’s fitting, because we think Spring one of the best.  Again, we admit that the fact that Spring Tide conformed to the format of our original VPN architecture proposal (September, 1999) made it easier to judge them on the points.  Still, this is excellent work.

As is our custom, we’re presenting the submission as-is, except for formatting, and coloring the text blue to distinguish their response from our comments.  The figures are provided at the end of the issue, formatted one figure to a page.  As usual, this is for subscribers only!

Down the Line

We’re still watching the developments in the Asub space, and we hope to cover the detailed plans of BANDI and ASI when the regulatory dust settles.  Next month, though, we’re going to review standards and topics in new-generation voice services.