Netwatcher (ISSN 0890-5800) is a monthly publication of CIMI Corporation. Subscription information is available here. Copyright © 1999, CIMI Corporation. All rights reserved. No publication or reproduction of this document is permitted without the express written consent of CIMI Corporation.

Management Briefing

Mom’s apple pie, the flag, and high-speed Internet for everyone.  It’s hard to argue with any of those icons, and certainly SBC didn’t attempt to in its recent announcement on high-speed services.  By 2002, they have DSL available to 77 million users (“Americans” in the press release, hence the flag).  Video mail is a service SBC likes for the future, a service that will let you at least see Mom’s apple pie, if not consume it.

The details of this announcement are even more interesting, and perhaps most interesting of all is how the whole thing has been covered.

SBC said that it will be spending $6 billion on the “Project Pronto” initiative to create a platform for the delivery of next-generation broadband-powered services, including Internet access.  About a quarter of this will focus on the office-to-office or trunking network, and the remainder at the customer edge.  The empowered customers will have what SBC calls “e-tone” (a kind of dialtone-on-steroids) to reflect the greater capability of broadband services to deliver versatile information offerings to buyers.

A large part of the press release, of course, is focused on the Internet and how DSL will revolutionize that experience.  Since SBC has, in the past, shown itself to be a canny planner and competitor, we attribute this focus to its recognition that only the “I-word” gets press.  Why do a press release if not to do that, so Internet focus it is.

The meat of the SBC announcement is a concept they call the “neighborhood gateway”, a broadband fiber remote that would replace the current GR-303 or DLC remote technologies.  SBC says they will push CO facilities closer to the customer, another way of saying that they’ll be smart devices with more than just traffic multiplexing capability.  SBC says they’re going to have 25,000 of these in place.

What SBC is announcing, of course, is a new-generation outside plant architecture for broadband service delivery.  They’re doing this just weeks after the FCC issued a critical ruling on packet wholesaling, and that’s no surprise either.

The FCC ruling arises out of the old appeals of the RBOCs to have their advanced services infrastructure exempt from the wholesale requirements of the Telecom Act.  The FCC ruled last year that they couldn’t do that, but promised additional rulemakings to guide the RBOCs in how to divide advanced services up between their ILEC core entity (which is subject to wholesale restrictions) and CLEC-like subsidiaries that would have no such obligation.

In September, the FCC issued a summary of its position on a key issue—just what the RBOC obligations for wholesaling digital next-generation infrastructure would be.  The Commission indicated that while the RBOCs would have to wholesale connection services out of a packet outside plant network to all competitors, it would not have to wholesale UNEs out of that plant (in general; there were some limits here that aren’t important for this summary).

This is an absolutely critical move, because the UNE issue on new digital infrastructure was the most troubling to the RBOCs.  Any broadband access network would have to depend heavily on fiber, and if they were obliged to unbundle the fiber and resell it as a UNE, the low price of residential services would set a low value on the glass, opening up a major competitive risk.  Now, the FCC has said that that unbundling doesn’t happen, and the RBOCs are thus free to deploy digital infrastructure out of their incumbent entity.

SBC has proposed that all along, of course.  Bell Atlantic, arch-rival, has yet to announce it’s own plan, but the FCC position certainly opens the likelihood that they’ll follow suit and deploy DSL and packet infrastructure out of the ILEC.

What’s really interesting about this announcement is that SBC says it’s going to be based on, guess what, ATM!  The requirement that the incumbents wholesale at least the total access service package to competitors is enough to push them away from IP infrastructure, since that would have direct resale appeal.

Look at the announcement another way, sports fans.  SBC is announcing that 77 million customers are going to be touched by a new ATM-based digital infrastructure.  Furthermore, via the Williams partnership SBC references in its release, that ATM infrastructure will extend over the long haul.  Between major metro areas, at least, we’ll have ATM capability.

The initial offering will involve a DSL-integrated primary voice connection and high-speed data connection suitable for Internet access or other applications.  Future offerings will expand to support up to 16 voice lines over SDSL to businesses, and four additional lines to residences.  Future offerings will also include support for SVCs to facilitate redirecting the Internet access connection to a corporate VPN for telecommuting.

SBC indicated that it is already committed to a significant xDSL deployment in support of telecommuting, which is no surprise given that this market offers higher willingness to pay and easier sales cycles than standard Internet access.  The telecommuting focus is consistent with what we believe is a business-directed sales strategy for the broadband services created out of this new structure.

Residential DSL access to the Internet isn’t a super revenue opportunity for the RBOCs in general.  It’s true that there are a zillion residential users, but it’s also true that there’s enormous downward price pressure on residential services of any sort.  The revenue gains SBC could expect from broadband access to residential users would probably be significant only if SBC could capture a significant number of the subscribers for its own Internet service, or if another type of service were to be marketed to residential users.

There seems to be no question that the Telecom Act and FCC regulations compel SBC to offer equal access to their DSL infrastructure, not to mention deals already struck with people like AOL.  There also seems to be no question that while there are other things that could be done with a DSL network to support residential users, none of those things show signs of creating a near-term market.  To earn a return on the early roll-out, SBC would have to rely on business sales.

There’s no reason why they can’t.  Neighborhood gateways don’t have to serve exclusively residential neighborhoods, and in the metro areas where SBC proposes to deploy first, many businesses are intermingled with residential units in any case.

OK, analysis time.  Here are the key points we take from the SBC release:

·         RBOCs will use the FCC packet-unbundling exemption to justify their deployment of digital outside plant and ATM interoffice networking technology.  They probably won’t spin DSL into a CLEC subsidiary, but rather will spin off data service provisioning.

·         Digital/packet outside plant will provide welcome relief to service providers from the congestion-induced limits to exploiting customer opportunity.  This should expand interest in new-generation services like VPNs.

·         New services over broadband infrastructure are mandatory now that the RBOCs have committed to lowering per-bit access costs.  Otherwise, their revenue model will tank to the extent that they are successful in the packet space.  Not a happy prospect.

·         IXCs will use ATM integrated access from the RBOCs to support their own integrated services initiatives.  Sprint’s ION, troubled by poor acceptance, can now at least be framed in a logical structure.  We believe that the IXCs may then “backfill” their long-haul services with local voice based on a feature-flexible architecture and compete with the ILECs.

·         Data CLECs will have a challenge building a business case on integrated access to the Internet, because the ISPs they support will be able to wholesale ATM-based access from the ILECs themselves.  Since packet UNEs won’t be wholesaled, this would mean that except for those users with home-run copper, data CLEC exploitation of the DSL opportunity would involve wholesaling the very service the ISP could get directly, then marking it up!

·         Voice over IP is dead in the local exchange because if the RBOCs don’t provision it, nobody of consequence will.  Thus, VoIP is dead in an equipment opportunity sense, since outside the local exchange the number of devices needed to support a VoIP network is just too small to fulfill companies’ earnings needs.  Hey, if this surprises you, you’ve never read any of our views before!

·         The FCC has decided that “local exchange competition” means ILECs versus cable companies, because the CLECs have failed to produce a rational business model despite having been given years of opportunity to run rampant when the ILECs were in regulatory never-never-land.  Continued propping up of the CLECs could only cause a delay in broadband gratification overall.

·         Virtually every service access device vendor has been obsoleted by the decision to re-architect the outside plant into multiplexed digital technology mode.  The products of today assume a physical media or TDM aggregation.  A new model of how services (including voice) will be offered to broadband customers is needed.

·         New-generation voice players need to focus on ATM access since that’s what will prevail on a large scale.  Many have developed product architectures which, if not IP-specific, are at least dependent on some deployment of IP to carry signaling.  At the very least, this should elevate the current “features-creation” players to the head of the class, since everyone else was presuming IP-driven evolution would eventually change the voice landscape.

·         Access infrastructure changes like the SBC one are more important in a revenue sense than IP in the near term, at least.  SBC will spend about a year’s worth of router sales on this modernization alone, not counting CPE, switching changes, and other good stuff.  They represent about a quarter of the likely local exchange refit spending.  A vendor play based on filling the need for big routers is, at the very least, aiming at “where they’re not” in a marketing sense.  Until access changes, demand for larger core boxes can’t develop.

·         Optical network evolution will likely be faster than electrical evolution inside the carrier cloud.  As is the case with edge networking, optical core networking can be seen as a positioning issue.  Carriers need infrastructure to sell anything to anybody.  That should help out incumbents Lucent and Nortel, newcomers like Sycamore, and validate purchase of Cerant and Monterey by Cisco.  For Cisco, it says that their hope of selling tons of big iron to facility-based carriers will be delayed behind the positioning deployments in the core and at the edge.

·         The press is clueless in this absolutely pivotal event.  The FCC ruling wasn’t covered at all, and the coverage of the SBC announcement not only omitted any insights, it also omitted most of the facts in the release.  One national rag did about a third of a page on the release and never even pointed out that this was an ATM play

There is always a chance that the SBC moves are simply hype, of course, but we don’t think so.  The RBOCs know that they need to support evolving broadband demand or be bypassed with respect to all the key customers.  This move clearly addresses that in a way that’s optimal, given the business and regulatory conditions that prevail.  What else, then, could they do?

In the Know

The nice thing about reality is that it happens, eventually.  The not-so-nice thing about reality in our industry is that it doesn’t happen easily.

There’s a ton of talk about the network of the future, and in particular about how that network will generate quintillobits of traffic.  This, not surprisingly, justifies huge switching capacities on routers or other information-handlers close to the edge of the network.

Well, it’s not reality—at least not any time soon.  The real requirements for traffic-handling in the network of the future aren’t all that hard to figure out.  The result of the process may surprise you.

Strategies

The Internet is always presented as a revolution in networking.  While we strongly disagree with those who think it’s the next public network model, it’s clear that it’s the next public commerce model.

We talk about how the Internet changes our lives, not realizing that most of the reason is that shopping for stuff is a big part of our lives, and shopping is what the Internet supports.

Business, of course, is somewhat dependent on the shopping paradigm for sales.  Thus, business is likely to support the Internet shopping process to the extent that it changes the way that opportunities can be realized.  To the extent that it does, the Internet will change business.

You see that proposition today in Nortel’s commercials, but we think that the business of the future is going to have to come up with a more compelling approach to its new market than acquiring a set of “juju eyeballs” to look at it through.  So what might that new approach be, particularly in the networking sense?

Two Faces of E-Commerce

Companies today tell us that there are two major ways that the Internet impacts the commercial world.  One is in the stimulation of purchasing and the steering of it toward a particular source, and the other is in the creation of an alternate channel for execution of the retail transaction.

In today’s market, most goods are sold in a two-channel model.  The seller advertises the product in a mass-media channel to create consumer desire and to bias the purchase decision in their direction.  “You need Widget Strawberry Jam!” is both an attempt to get the viewer/listener to want a jelly sandwich made with strawberry jam, and to think of Widget when satisfying that craving.  The seller then provides places (called “stores” for those who don’t get out much) where the goods can be acquired.  The idea is to get the consumer from couch-potato mode to a shopping mall, wallet in hand.

Perhaps the greatest problem that traditional advertising has faced in exploiting opportunity is the problem of “targeting”.  An ad for strawberry jam on the evening news is a good play if the desire for strawberry jam can be created at that time, and if the desire so created can be sustained until the purchase is made.  A lot of people will see the evening news before dinner, a meal at which strawberry jam usually isn’t featured prominently.  Few people will probably develop much interest in it at that time.  Fewer will remember the desire when, in the future, they happen to be at the store.  Bad targeting.

The Internet, unlike mass media, offers what might be called “invited advertising”.  A user with a desire to buy something can run to the Internet and find out how to gratify that desire.  Instead of trying to dynamite a bunch of half-asleep consumers out of their chairs and herding them to the mall, the Internet gets the consumers who are already of a mind to shop.

A web site is an invited ad.  People aren’t steered to your website by the cosmic forces, they get there because they clicked through the necessary URLs.  If your content satisfies their requirements to determine if you’re a credible source for the service or product they desire, they are disposed toward you in their purchase decision.  Since they had to be surfing for a reason, it’s more likely that they’re really trying to make a purchase decision than it would be for the “watcher” of a TV to be contemplating a purchase.

This explains why new research is showing that “injection ads” on the Internet aren’t particularly effective.  The web-surfer isn’t interested in those ads any more than the TV viewer would be interested in the mass-market alternative.  The difference is that the TV viewer is a captive to the fact that the TV shows what the station broadcasts, and their favorite sit-com is going off the air until the ad is finished.  Faced with the alternative of watching the ad and doing something conversational or socially useful, the tuned-in generation watches the ad.  In contrast, the surfers don’t see the ad at all, focusing as they are on the content they’ve sought.  One Internet ad in a hundred is even noticed, which is why the current trend is to make them look like popup windows.

OK, the consumer is on a mission and the Internet has let that consumer get quick information to guide them.  That’s where purchasing comes in.  If the item isn’t an instant-gratification item, or if the item isn’t one that the consumer requires personal experience to purchase, it may be possible to convert the interest that the Internet invited ad can create into an instant gratification.  After all, ad research has always shown that any delay between the development of a need and the opportunity for the gratification of that need leads to a reduced chance the user will pursue the necessary steps to execute.

Today, our research indicates that the Internet’s influence on commerce is overwhelmingly in the “ad” mode, not the execute mode.  Over time, however, there is little doubt that the proportion of executes will increase.

The Impacts

Getting on the web is the thing that will result in the greatest overall benefit gain for businesses addressing e-commerce opportunities.  The reason is that the retail infrastructure we have today gives us an option to present ourselves where the product is sold and take delivery.  That takes less time than having the product shipped.  Instant gratification means gratification in the form of receiving the product, not just in completing the sales execute (except, of course, in areas where there’s normally a delay between execute and delivery, as with a car or house).

For companies that shift to web promotion, the issues divide between message coordination, message content, and the integration of their core systems to their websites.  All three must be matured if the Internet experience is to succeed.

How does a user get to your website?  Answer:  the user knows your URL or the user links from somewhere else.  To make the Internet experience successful, a company must insure that one or both of these avenues are created and maintained.  The nature of the way a particular seller would balance these channels would depend on the seller’s assumptions about the web-literacy and interest profiles of the buyers.  That’s not a technical issue, so we’ll leave it there.

Marketing and sales are separated by the concept of immediacy.  A web site is more a selling tool than a TV ad, because it is presumably hitting someone who has a much higher likelihood of having decided to purchase something already.  So, the material on the web site is targeted at selling the sales contact, not the product family, as a general rule.

Where impact with existing systems comes in is where the credibility of the buyer wanting a sales contact will be based in part on the seller’s price, availability, or other variable factors.  A widget is a widget in a description sense, but if the buyer wants a widget at a given price, or wants assurances on widget availability, the seller will have to link the price/inventory status of widgets to the website to provide the necessary stroking.

The normal way to provide integrated product status and price information to a website is CGI or Java integration.  In either case, a portion of the HTML code invokes a data link to the business application that has the necessary information, and this information is integrated into the display the customer sees.  In both cases, the link is usually made using a TCP/IP connection.

Introduction of a TCP/IP link to a host application isn’t necessarily a network-shattering event.  Most users will have their MIS systems centrally located.  If the web servers are likewise centralized (and it’s hard to validate any other model), there can be a local link between the two to support information delivery.  Whatever application systems might of used the MIS data would be unchanged by this.

MIS transaction load would not be unchanged, clearly.  In fact, the biggest impact of web selling is on the computer resources of the seller, not on the network of the seller.  The current network stays as is, an Internet connection to the server is made, and the poor MIS data center system gets hit with activity from both networks.  That’s why IBM is probably a bigger winner in the Internet game, in the long run, than even Cisco.

This linkage is also where the impact of customer execute via the Internet would hit first.  A web retailer might start with a price/availability model of a web site, but would likely quickly transition to an execute model unless the product/service were such that only conventional retail fulfillment was logical.  Today, the majority of e-commerce companies who go beyond advertising product information on the web will move right to retail execute.

In the retail model, the transaction load on the host increases by the need to provide an update to the databases that were previously only used in inquiry mode.  There are also new requirements to insure that if stock status is a condition for accepting an order, concurrent ordering doesn’t end up committing more goods for sale than are actually available in inventory.  Again, these are really computational issues and not issues for the network.

Where network changes begin to occur is when the customer support process is introduced.  The customer has a view of the company based on the web.  The company’s own employees see the company in the form of a bunch of antiquated host transactions—the “green-screen-and-stickman” model.  Most companies have found that it’s difficult to support a customer who has a different system view than the customer support rep has.  Thus, it is likely that the transition to web marketing will involve a transition to web support.

Now, we have the classical ripple effect to consider.  Our CSR is supporting the customer with a web-based view of the company’s MIS process, at least insofar as that process is visible to the customer.  Will that CSR now shift to stickman mode for other MIS activities?   Even if there’s no such tendency, the CSR transactions will have moved off the legacy network and onto the IP network.  If the CSR is bitten by the browser bug, then all that CSR activity transitions to IP.

Now, we’ve got a problem.  CSRs inquire about inventory, but so do purchasing agents and others.  Do we maintain two versions of the transactions, one for the web-enabled and the other for the stickpeople?  If not, CSR-related transactions move en masse to the IP network, whoever invokes them.  Those users are now faced with what was the original CSR dilemma of whether to be part web and part stick, or go web entirely.

But getting these new transactions on IP isn’t the same as getting them on the Internet.  The Internet was the input conduit for the transactions from the outside.  In semi-technical terms, we put a CGI gateway on an internal LAN, linked it to the host via TCP/IP, and then provided that internal LAN with a firewall connection to the Internet.  If we wanted to provide our internal users with similar access, they’d have their own connection to the internal LAN.

Conclusion

This is the model that will drive most e-commerce-based network evolution.  It’s a model that relies less on the traffic of the Internet than on the behavior of the workers.  It’s a model that clearly separates the process of Internet access from the process of IP access for the new host applications.

The internal transactions associated with retail fulfillment have about 20 times the traffic volume of the retail executes.  It’s not clear whether those transactions have to go onto any WAN; that depends on where the internal users of the data happen to be with respect to the MIS data center.  It is clear that they’d transition eventually to IP whatever the location.

Not all businesses will be driven by e-commerce.  Not all who are will ever have a viable model for retail execute online.  Those who do are facing an inevitable migration to TCP/IP-based access to their MIS data.

The execution of an IP migration could take many forms, but it should be clear that the big question that will face users contemplating such a migration will be the “in-house-or-outsource” decision.  Many people with private networks today don’t think that they’d do it again if they had the choice.  Many now will have that choice—in theory.

The question will be the credibility of the public IP services at the time the buyer is making the decision.  As we’ve said many times, the Internet is itself not a credible way of creating a public IP service to support internal company communications.  That’s particularly true if we consider that the current internal network is SNA—about as far from the Internet as you get, culture-wise.

Down the Line

Next month, we’ll see if we’ve gotten enough responses to our VPN inquiry to provide you some early results.  We’re also looking to include a more detailed view of at least one XML voice player.  December is our Annual Technology Forecast issue and, as usual, won’t be published on the Internet at all.