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.

We’d like to open this year with an overview of what we think the key issues will be for the coming year, and to a degree for the new century. In this presentation, we’ll focus on the issues and not the answers (yes, since we’re consultants, you’ll have to pay for those).

The primary issue for this year will be the question of public data network services. It is clear from the success of the Internet that a ubiquitous public network can change the way business operates. It is clear that the potential new ways we could do business are exciting to stock markets. It is less clear that we understand how the limited success of the Internet can be leveraged further.

Limited success? Heresy? Well, total ISP revenues are about a half of one percent of service provider revenues overall. Total e-commerce value is about the same as the annual sales of one big corporation. Might it grow astronomically in the future? Sure, and that’s what stock players are relying on. Will it? Who knows, and that’s the problem.

We have surrounded the Internet with such a web of fantasy that it’s hard to view any aspect of it objectively. That makes it very hard to decide whether the technology of the Internet, the business structure of the Internet, or the free connectivity paradigm the Internet represents are readily translated into a broader service mission.

No matter what people think, the great majority of data activity is internal to companies. We estimate that it probably takes 22 internal transactions to fulfill a single external one. A public data service clearly has to address the internal traffic of businesses to be enormously successful, or it has to rely on a new source of demand and revenue.

If we’re to handle internal transactions with public data services, then how many of the Internet’s properties do we need to retain? There are millions of hosts on the Internet, but the average multi-site business has less than fifty sites. Why do they need an architecture that scales to the universe? They don’t, obviously.

The question is then whether an Internet-like architecture is necessary to provide the VPN features such a buyer would want. We can build a VPN with a small number of sites using a pure circuit model. We can build one using a pure connectionless model as well, and any number of hybrids in between. Which is best? That’s a good question, but we aren’t working to answer it.

The business question is just as troubling. Universal access to the Internet for effective e-commerce will clearly require subsidization of buyers by prospective sellers. We watch network TV today because an advertiser is paying for us to do so. Any marketing forum will work the same way.

This presents problems when there are proposals to use the Internet for something other than commerce. How much will "advertisers" on the Internet pay for chat rooms, or e-mail, or (more to the point) intra-company traffic of competitors? Zero, if they have any choice. Yet if we assume that "the Internet" is how these things are provided, they are supported by the same subsidization process.

The second problem we face this year, and into the future, is the "service/bandwidth" problem. The revenue stream for service providers has been traditionally linked explicitly or implicitly to bits for bucks—capacity is the stock-in-trade. That conflicts with both the trend toward low-cost bandwidth through technology innovation, and the desire of buyers to have higher-level services.

If carriers sell bits, then they see lower profits as bits get cheaper, unless somehow we consume a lot more bits in the future. That consumption is linked to the creation of services that consume more bits than voice and data, however. If those services are offered by the same providers who sell the bits, we have a simple transfer of revenue base, and no big problems occur.

If others, specialized providers of service, buy bulk bandwidth and translate it into services, then the margins of those players rise and those of the facility players fall. So what, you think? So the facility people are the ones who provide the real network, without which we not only don’t have the new services, we don’t have the current ones either.

Telecom reform is clearly disappointing a lot of people, but most of those who feel out-of-sorts with it have no real recourse. A few, representing businesses who would profit from being on the service side of the service/bandwidth issue, have big bucks to lobby. We may end up with political intervention in a matter that politicians know little or nothing about (OK, what other kind of matter is there, but we’re making a point here).

The third issue is the sea change in network and technology procurement. Nowhere in industrial civilization is there a market where technical planners have less to say about what technology is bought and used. Some will say that the populism, in business terms, of technology has brought us tremendous productivity benefits, and they are right. Burning down forests also improves land production, for a time. Eventually, conditions change.

Computing and information policy isn’t something you can decide or create by acclimation. Somebody has to insure that all the pieces of network technology that anyone can buy these days will somehow combine to form a useful network. If there is no coordination in purchasing, then we’re relying on vendors and standards bodies to do this. The latter group is too slow, and the former group probably has other priorities—like selling their stuff instead of having the competition make the sale.

In the LAN space, we can see that cheap switching is already making it difficult for the traditional big players to maintain direct account presence. Without direct sales presence to funnel strategic recommendations through, sellers really have no influence on policy, only one on product.

The same thing happened in the computing space when mainframes were threatened by minis, and IBM’s effective monopoly was broken. Yes, we ended up with cheaper computers, but we went through a period of turmoil and excessive IT spending, a period of compatibility problems and re-executed applications. All that happened when information technology use was in its infancy. What would happen today?

The final issue is really a hard one. Every year, buyers tell us that they are less comfortable with their vendors, their consultants, their reporters, their editors. Every year, we have as an industry deviated further from the path of reality to pursue commercial gain. Market myths today carry more weight than the truth, and the real world can’t be created or sustained by myths.

Each of the issues we’ve noted above are important to us all, and yet each of them has been around for at least three years. That we have no real resolution to any, in fact little progress toward resolution, can only mean that we have not accepted their importance. Why? Because no one has found it beneficial to deal with the issues, and no one has called anyone to task for failing to do so.

"VPN Products Confront FUD" according to a recent article. They sure do, not in the least because there’s no accepted definition of what a VPN is. The vendors that are discussed in the article don’t even have a consistent view, but that point is never made. Many would say that this is because the editorial slant of a publication on an issue tends to remain fixed once set, no matter how the issue evolves. Some would say that little real investigation or reasoning goes into creating that initial setting. But why do incomplete or incorrect positions remain unchallenged for so long?

Some of the problems are created by our litigious nature. Nobody (not even us) will stand up and say "This product absolutely stinks", or "That statement is absolutely a lie" because of a fear of legal process that could be so costly it would be irrelevant whether you won or lost. But there are ways of questioning claims that don’t inject legal liability.

Ultimately, the solution to the problem of truth (or lack of it) must come from the listeners. If you pick up a magazine with vacuous junk all through it, don’t accept the junk because the rag was free to start off with. Write the editor and complain. If you hear a claim, review an architecture, that makes no sense, complain about that, too. A discriminating consumer of information promotes value, and lack of discrimination promotes nonsense.

For anyone but the extremely uninformed, the acquisition of Ascend by Lucent was no surprise. Rumors on this topic have abounded for a year, and the move was clearly logical, and favored by strong constituencies on both sides.

Prior to the Ascend acquisition, Lucent was the king of the traditional service provider market in the US by a considerable margin. They also have a big chunk of the premises voice market (PBXs, key telephones, etc.). But while they had LAN products, a router, ATM gear (acquired with Yuri and some internally developed), and nifty multimedia stuff, they were not really competitive with Cisco in the data space.

Cisco, a sixth or so Lucent’s size in revenues, had pretty much the opposite situation. Cisco is the dominant player in the enterprise data WAN product market space, which is the only element of the data market that has been sustaining significant sales and margins at the same time. Cisco’s problem is that it has little penetration in the service provider space. The routers it sells there are resold to enterprises. It has some recent ATM switch success there, but not enough to create tens of billions in revenues. It has no market share in voice.

From these brief descriptions, it should be clear that the two giants have not really been boxing, or even sparring, but rather throwing punches at a 90-degree angle to each other. Cisco threw a weak jab with Summa Four. Lucent parried with a soft PacketStar router. Nobody laid a glove on anyone else. Instead, both firms have been following what might be called a political course.

Cisco’s Chambers recommends throwing out all those antiquated PBXs. He touts IP voice as the way of the future. He presents the telephony world as only slightly more modern than the late Triassic period.

Lucent kisses the IP voice baby. Lucent endorses IP on the premises, and integrated messaging. Lucent meanwhile buys a big ACD competitor, and continues to rake in Class 5 switch dollars to earn record levels of bucks. IP is great, TDM is great, voice is something most of us have, and money is something we can all respect.

Cisco is winning the war of words, because the media loves revolution and Cisco is promising that. Lucent is winning the war of dollars, and that’s where true victories are gained. But could PR victory in a market lacking in insight translate to real gains for Cisco? Could it be that enough headlines on IP convergence might eventually lead even reasoning buyers (who, heaven love them, are certainly an oppressed minority if ever there was one) to IP, all truth and knowledge notwithstanding?

With the acquisition of Ascend, Lucent is committed to selling ATM switching products, and specifically ATM switching products with a mission to do IP-over-ATM, to service providers. That means that Lucent is creating a beachhead in the data space in the IP/ATM area.

That’s the same area that Cisco has been staking out as its way to jump out of being a provider of routers to carriers (as an enterprise channel) to being a provider of switches to carriers to install in their own network. Cisco knows that the big facility-based players have all the money, and that these guys won’t invest in pure connectionless architecture in the near term, no matter what the market says.

What we have here is super-power number one, who’s been eating up little countries starting with the tip of Africa, and super-power two, who has been chomping west from the Bering Straits toward Land’s End, coming together about Switzerland. They may have started off at right angles, but they’ve converged on a common market.

Now they have to fight for real.

Cisco needs to sell a ton of ATM and IP products to the RBOCs as they build out their infrastructures to prepare for national, multi-service, competition. Why? Because that build-out will involve a minimum of $150 billion over the next decade. Any year’s worth of it is more than Cisco’s current revenue stream.

OK, Lucent, you bought Ascend. What do you do with it? The answer, in short, is to do what Ascend was being pretty successful doing at the product conception and market conception level, but better than what Ascend was doing at the sales level.

Ascend has a strong incumbent position with virtually all of the RBOC kingpin players. Their ATM gear is installed and earning revenue. This equipment can be quickly put into IP-over-ATM service and address a market that would be ideal for the RBOCs to use as a jumping-off place into competition with the IXCs—the VPN market. Ascend didn’t have a lot of the ancillary gear, including premises access. Ascend didn’t have the image of solidity. Ascend didn’t have sales account control even though they were incumbent. Lucent has all those things.

What Lucent doesn’t have at the moment is a real position in the whole IP/ATM debate. Lucent has fielded a router (PacketStar), and it has MPLS capability. Lucent has purchased Yuri, one of the most vocal supporters of the CSI initiative for IP-over-ATM based on MPOA. Lucent has now acquired Ascend, whose IP Navigator is based on MPLS principles, and is or will be MPLS-interoperable. In effect, Lucent has a foot in all the camps.

Can Lucent simply adopt the Ascend position? In some ways that would make the most sense for them. Ascend’s equipment is incumbent, after all. Ascend’s solution is already delivered and publicized, and an alternative strategy would clearly leave Cisco romping happy and free for some time while Lucent put the pieces into place.

Ascend’s approach is also a good one. While one could argue how scalable IP Navigator is (and Cisco will certainly do that), the truth is that it’s more scalable than most VPN opportunities would require. The critics of Ascend start with the assumption that VPNs are a subset of the Internet, and that IP Navigator can’t support the Internet. The first is false, so the truth of the second is irrelevant.

But despite the value of IP Navigator, it is probably true that the concept needs some expansion. There are a number of models for MPLS or virtual circuit-based VPNs emerging (see Strategies in this issue—if you’re a subscriber and not reading this on the Internet) and IP Navigator doesn’t support them all. It seems likely that Cisco will.

A second option for Lucent would be to try to pull Ascend into its own positioning for IP. The problem with that approach is that nobody (except perhaps within Lucent) knows what that positioning is. The CSI stuff seems to be dying a deserved death as an alternative to MPLS, but Lucent ironically seems to contain more CSI holdouts than Newbridge, who invented the concept to start with, and who has already positioned it as a supplementary approach to MPLS (which, if CSI is anything at all, is true).

A non-religious IP/ATM position might make some sense market-wise, but it’s hard to see how the company would retain PR control and control technical scope with such a position. The key would be for Lucent to announce a really far-reaching and compelling architecture for the whole 21^st-century networking shebang, and make a killer IP/ATM strategy a part of it. In this strategy, they could position both PacketStar and CSI as supplementing the IP Navigator base approach, and the result could be something that would not only be a killer concept, but more readily understood by users.

A face-off between Lucent and Cisco is a kind of super-power nuclear exchange, but just as in real wars, it’s often the peripheral players who pay the greatest price.

Nortel is hit by the deal at its weakest point and perhaps weakest time. This fall, Bay’s House is expected to depart, the man who finally succeeded at unifying and motivating Bay’s fragmented product groups. Nortel has yet to create a vision of a unified company, their allegations notwithstanding. Whether they can do so now, and implement the vision if they do, is a question.

They need to do something, because the Cisco/Lucent struggle will focus buyers on the very things that Nortel has failed to come to grips with so far. Is the future ATM, IP, MPLS, SONET, DWDM, or what? Nortel, more even than Lucent, has covered all the bases. In this PR-driven market, taking all possible stands is taking no stand at all.

Newbridge is even harder hit. A side-by-side competitor with Ascend in most of the former’s RBOC accounts, Newbridge has survived its seemingly complete loss of market vision in the last year only because Ascend’s sales and marketing efforts to the RBOCs haven’t set records either. With Lucent taking over, and recognizing that the RBOC accounts are where they must confront Cisco if they are to make the Ascend buy pay off, Newbridge is camping in Switzerland (to continue our prior analogy) when the warring armies converge there.

Newbridge does seem to have an IP/ATM vision in a marriage of CSI and MPLS, but they appear to have lost the ability to promote that vision effectively at either the marketing or sales level. That problem could be corrected more easily than a real product confusion or product lapse, but the company has been paralyzed in the market positioning sense since late 1997, so it’s hard to see why they would break out now—or how they would do so.

Even if they do break out of the IP/ATM doldrums, they face a problem with breadth of product line. Just as Cisco must inevitably confront Lucent in the Class 5 space, so must Newbridge if they want to stay competitive. They have no products there, so it would mean an acquisition. This isn’t the most favorable time for one, of course.

A final problem is the Newbridge/Siemens/3Com alliance. Clearly that alliance could bring Newbridge everything they need in a product sense. Clearly Newbridge’s Lutz sees that; his speeches have been almost fawningly supporting of Siemens in particular. Clearly Siemens isn’t all that impressed. 3Com’s perverse pressure on the concept of an IP-based PSTN, which plays against the strengths of either of its partners and into Cisco’s hands, puts pressure on the alliance overall, making it more likely the whole thing will tank.

But how about the main event? In our view, the real question for the next couple of years, at least, is not who does the best IP/ATM but whether Cisco can successfully attack the Class 5 and PBX market. Lucent needs to keep Cisco from establishing a foothold in the RBOCs using MPLS, but that foothold couldn’t threaten Lucent as long as Cisco continues to rant about IP telephony as its only voice strategy. Long before they could make any headway, Cisco’s enterprise business would drop off, and they’d fall into Cabletron-like stasis.

This isn’t to say that the Ascend deal won’t be significant. By moving into that space, Lucent can cut Cisco off from what would be a significant near-term revenue stream and hasten the time when Cisco’s dependence on enterprise WAN IP will compromise its revenues. That will shrink the window in which Cisco can prepare an effective telephony position.

It’s also true that if Lucent blows this deal, they may hand Cisco the keys to the kingdom. A prompt and rational telephony strategy could well grow out of Cisco’s new IP-ATM vigor, and if ATM positioning ends up hurting Lucent because they hit a pop fly with Ascend instead of a home run (so to speak), Cisco could claim a big chunk of the RBOC build-out, and overtake Lucent in revenues.

In the recent announcement of "IP-Enabled Frame Relay" by AT&T, even the lack of technology detail can’t hide the fact that there’s an issue coming forward in the VPN space. In fact, it may be the pivotal issue of the VPN market, now that MPLS has earned a position as at least the PR foundation of future VPN services from facility carriers.

VPNs, as we’ve said a zillion times, are a marriage of IP flexibility and virtual circuit performance and determinism. Given that, it might be reasonable to assume that VPN architectures are a marriage of connectionlessness and connection-orientedness. It might also be reasonable to assume that there are a number of ways this could be done, and a number of issues that might influence which of these multiple approaches would serve best.

In our next issue, we’ll be looking at the increasingly hot "next-generation-voice-infrastructure" space, one that is attracting a number of new players. We’ll also review some key regulatory trends, ones that could impact how we all consume networking for decades to come.

Access the index of CIMI Corporation's recent newsletters.