Engineering Crossroads March 1, 1998 Volume 28 Issue 1 The Bridge, Volume 28, Number 1 - Spring 1998 POTS and PANS: Telecommunications in Transition Sunday, March 1, 1998 Author: George H. Heilmeier Competition, customer-related concerns, and cost are forcing a major shift in the telecommunications industry, with significant implications for the engineering profession. I am going to talk to you this morning about POTS and PANS. I think some of you who've been around the telecommunica- tions industry know what POTS stands for: plain old telephone service. It has been the bread and butter of the local exchange carriers for many years. Fewer of you know what PANS means in this context. By PANS, we mean "pretty awesome new services," and that's exactly where this field is headed. In my view, there are three factors forcing transitions in telecommunications: competition, customer-related concerns, and cost. Let's consider competition first. Technologists always look at technology being at the focal point of change, and it certainly is. It was the convergence of computing, communication, and services that really began to drive the most recent major changes in this industry. That convergence essentially led to deregulation, and deregulation led to domestic as well as international competition. And, yes, it led to some teaming, which in some cases seemed an unnatural act. Another result of deregulation was that size really became important, diversity became important, and so, over the last several years, we've seen megamergers in our industry. Finally, we've seen some major changes in the way business is conducted. Large telecommunications network owners were suddenly finding themselves in a world where they really did have to compete, and, let's face it, competition is only just beginning. No More Captive Customers The second forcing factor is the consumer or customer-related trends. It used to be that all we had to worry about in this business was POTS. It's not so simple any more. Now, we have to worry about video, electronic commerce, entertainment, information access, the Internet and intranets, and the work-at-home desire. Commercial customers and customers in the mass market are making choices just the way people make them in every other industry. They look at features, they look at price, they look at service, they look at access. They also weigh things like pervasiveness, and they want a common look and feel. They expect interoperability and universal connectivity. There are no more captive customers anymore; we can't take the customer for granted. This has really been tough on the old monopoly stakeholders, because they have had to learn a whole new way of doing business. It will be very interesting to see how long it will take for a major Bell company to bring in as a CEO someone from outside their own system. I came from outside the Bell system, and I can tell you that even for someone at my level, it was difficult for many people to accept. I can only wonder at how difficult it's going to be facing new challenges and new markets for a major company to acquire a CEO with experience and skills aligned with the new competitive realities. The third forcing factor is cost. It used to be that when people talked about costs in the telecommunications industry, they talked about operating costs and capital expenditures. Those are still important. But there's a whole new dimension now to what costs mean in this industry. In addition to labor, facilities, and maintenance, people now look at barriers to value-added activity and what these are going to cost them. They are more sensitive to faulty quality, to equipment downtime, and to obstacles to innovation. They also deal with a problem common to most companies with large legacy systems: Most of them are "stovepipe systems" where the component units don't talk to one another. The result can be a horribly inefficient process. For every point of flow-through that's missed, it costs a regional Bell operating company anywhere from $5 million to $10 million. So, the ability to integrate, or harmonize, these systems is important not only from the standpoint of providing flexibility and new services, but it's absolutely critical to driving costs down. What I'd like to do now is look at how these forcing factors are leading to transitions in the industry. There are transitions in the industry itself, in the networks, in the services offered by the industry, and in software, which is really the glue and the brains of the network operators. Today, the telecommunications sector is beginning to reshape itself, from a vertically to a horizontally structured industry. That's exactly what happened in the computer industry almost a decade ago. The process is still under way in our industry, and the transition is very difficult for some people to accept. These companies were not organized originally to support a horizontal structure, so the shift has created enormous technological and social pressure. Now, the question of efficiency focuses on how well we do something in-house versus how efficient we could be if we took advantage - as the computer industry has - of other specialized players and their services. Furthermore, it used to be that new capabilities were driven primarily by the carriers. Now, they are beginning to be driven by the users. That's what the advanced intelligent network (AIN) was all about. In the past, the carriers were responsible for new capability because services were very closely bundled with switching. In order to introduce a new service, you had to go to your switch manufacturer and argue why it was a good idea to open up somewhere in the vicinity of 10 million lines of code in order to gain the ability to support a new service. That, in part, is why it took so long for new services to be introduced. What AIN enabled was the unbundling of service creation and service operation from switching. Many people thought this was going to revolutionize in the industry. We thought it was going to open up service creation to third parties and clients. But AIN never realized its full potential. Why? Many saw the opportunity with AIN to once again learn some lessons from the computer industry, in particular the personal computer industry. We saw that no successful PC company was also successful in introducing blockbuster application programs. No matter how hard they tried to incentivize their engineers, they never came up with those killer application packages. Other folks did, third parties. There was de facto unbundling. Similarly, we in the telecommunications industry saw unbundling as an opportunity to construct an environment that would permit the creation of new services and applications. Well, it didn't exactly happen as planned. Network operators just didn't want one common service-creation environment. Each wanted its very own, thinking that this was a route to differentiation. People didn't recognize the lesson that the PC industry was trying to teach: There's a universe of people out there who have a much better idea than we do of what key applications are, so why not give those folks the opportunity to realize them? Two de facto operating system standards emerged. Well, it doesn't happen that way in the telecommunications industry. And that's why today we have POTS, the traditional role of major carriers, and PANS, which is still something we're working toward. These same forcing factors are also changing the telecommunications network itself, which is still primarily circuit switched. Soon, I think, the network may be relying to a considerable extent on the Internet protocol (IP) for practically every service. Will circuit switching go away? Not likely. But the IP is going to drive telecommunication networks in the future. For data and all kinds of customer-driven applications, IP technology will be extremely important. The expectation that IP telephony will be just as good from a quality standpoint as today's telephony will not be met, at least for a while. Among other challenges are those related to latency and delay. What could very well happen is that we'll introduce Internet telephony and people will be disappointed with the quality, no matter how cheap it is. We'll go through a stage that's very similar to what happened to cordless phones, namely, an initial explosion in the market followed by a large degree of customer dissatisfaction. When cordless technology finally did catch up, the consumer was very wary about reinvesting in the product. Now, that market is robust again because of what technology has made possible. The same scenario may very well play out with Internet telephony. One of the key questions that the journals love to write about is which access technology is going to win out. Today, we primarily have fixed-access technologies (e.g., the twisted pair), but there's also wireless, which is growing dramatically. In the future, access technologies will include fiber to the curb, fiber to the home, hybrid fiber-coax, ADSL (asymmetric digital subscriber loop), and wireless. It's my view that carriers are going to have to support all of the possible access technologies, because one size certainly will not fit all situations. We're also going to see major changes in the customer-premise technology, which today is the simple telephone. Because the telephone is such a simple device, most of the "smarts" are centralized in the network. We're now seeing a shift to intelligent customer-premise equipment like PCs. This raises the issue of where the smarts are going to be in the network of the future. I'll talk a bit about that later. Efficiency Up, Employment Down Telecommunications operations still rely pretty much on technicians, but there has been a rapid shift from routine testing to proactive surveillance and software-oriented operations. When the Bell system was broken up in 1984, an average regional Bell operating company had approximately 70 employees for every 10,000 lines. Today, that number is closer to 25, a reduction that has resulted from transition to a software-intensive network. In part, it's that transition that has enabled the regional Bell operating systems and others to operate much more efficiently than their counterparts in other regions of the world. Security and Robustness What about security and robustness? We expect both from our telecommunications network in this country. In many respects, we're the best in the world in these areas. But what happens as we move to an unbundled network? There will be major problems, in my view. In an unbundled world, people want access to key nodes in the network, and they want access to network databases. From a competitive standpoint, there's no reason why they shouldn't get that. From a standpoint of security and robustness, however, we should tread very, very carefully down that path. Unbundling, while it fosters competition, also raises some key issues. I don't want to raise the same arguments that were used back in 1984 by people who wanted to keep the Bell system together. Still, by giving people access to network databases and switches and the like, we're really treading into the unknown in so far as security and robustness are concerned. While I don't think the network is nearly as vulnerable as some of the "junk" scientists would have you believe, there are legitimate reasons to be concerned. From providing support primarily for voice transmission, networks are beginning to offer a raft of new services that require strong support for data as well as video. In the past, most services were discreet and were billed discreetly. Today, the key in an emerging competitive world is going to be service bundling. For example, if you buy a movie from me on a pay-per-view basis, I'd like the flexibility to give you credits for long-distance service downstream. That requires a billing system that doesn't exist today in the telecommunications industry to any great extent. So, the introduction of bundled services to enhance connectivity has a barrier problem. What's needed are flexible billing systems, systems that can handle real-time rating and discounting. What about software? Software drives the business today, yet once you get to the level of vice president and above, most folks don't really understand the importance of software to their companies. In the past, the complexity of software in major telecommunications networks was in the 10,000-function-point class, or systems containing on the order of 1 million lines of code. Although these are very complex systems, we are moving to even more complexity. But the new systems will be much more modular, flexible, and scaleable. Today, most large companies are grappling with the results of stovepipe systems, leftovers from the 1970s and 1980s. Back then, people automated specific tasks without paying much attention to overall business processes or the need for data and computation to flow through many different applications. Well, the watchword for the future is integrated systems. This transition is under way, but it is not going to occur overnight. It is going to take a while, because today the average regional Bell operating company is supported by over 125 million lines of code. Building good solid industrial-strength software of that size and complexity takes a lot of money. The architecture of the major telecommunications software systems is mostly tool centric. We have got to move much more quickly than most people realize to become data centric. In the past, practically every system was a customized job. The process was very expensive, very time consuming, and very error prone. Software systems of the future will probably be assembled from reusable modules rather than built from scratch. Another problem on the horizon is the issue of feature interaction. The more services that are created by different groups, the more significant this problem becomes. A major concern today is whether asynchronous transport mode (ATM) or IP will be the basis of the network platform of the future. For my money, it's not going to one or the other, it's going to be both. There are some who think that IP and Sonet will make up the entire network backbone. My own view is that there is going to be something in between IP and Sonet, and it's going to be ATM. Real-time services such as telephony and interactive video may very well force a harmonization between IP and ATM. What about the SS7 network, the backbone signaling network for today's circuit-switch networks? Well, I think IP routers will eventually replace the STPs, since the routers are an order of magnitude less expensive. But, in my view, it is going to take a decade for that to happen. Will Internet telephony transition from "toyware" to something real with significant market penetration? I think the answer to that question is, yes, but it will not happen nearly as rapidly as many people think. And what about the Internet model itself? Can it perform reliably, can it be industrial strength? Of course it can. But we will go through some shakedown cruises here as well. It is not going to happen overnight. I don't agree with those who predict that the Internet model will represent roughly 10 percent of the voice telephony market within 5 years. Certainly, the Internet model will be pervasive in networks of the future. So, where is the value-added going to lie in these networks? Today, as I noted earlier, it is centralized. This is because the customer-premise equipment is a telephone, an unsophisticated device. The smarts have to be buried in the "middleware" of the network itself. But that is going to change as more-capable user equipment (e.g., PCs and servers) is distributed throughout the network. When it does, the economics of this industry may also change. Those folks who are operating the circuit-switch networks today are going to have to learn a great deal about some technologies that are completely foreign to them. If they don't learn and learn very quickly, they are going to become dinosaurs. A technology of future importance is wavelength division multiplexing (WDM), which is beginning to be deployed in trunk networks to provide increased bandwidth and capacity. The real issue is when will we evolve into an optical network? Such a network will have not just WDM but also optical switching routines and a raft of necessary network functions. This could very well drive a new generation of operations support systems. How are we going to get broadband services into the home when the path to the home is twisted pair, a technology that goes back decades? For some, ADSL will be the answer, but it won't be pervasive. For one thing, ADSL doesn't fit some of the local plant that is out there today. Many existing plant surveys are outdated, hence you really can't tell what's in the local loop. In some cases, things were deployed a long time ago, were not documented properly, and are in essentially "write-only" data bases, so nobody can get to the data. Expensive on-site inspection may be required. A lot of things stand in the way here. You read that ADSL is the way that people are finally going to get broadband access to the Internet from their home, and I would love to believe that. But it is going to take a while. An additional challenge is harmonizing ADSL or the XDSL families of access devices with the existing system software infrastructure. Those software system modifications are just beginning to happen. Of course, even if you had a broadband pipe into the home, it does not necessarily mean that you are going to get high-speed access to data. Network operators are going to have to learn something about caching various network sites, they are going to have learn something about extranets in order to work the problem end to end. Carriers today are focusing mostly on bringing broadband capability from the central office to a residence or to a business, but that alone is not going to solve the problem of high-speed data access in most cases. Skills for the Engineer From the standpoint of the engineering profession, some key skills needed by those working in the telecommunications industry are not going to change very much. Traffic management, even in IP networks, will remain important. Experience in network architecture, network design, operations, and large-system analysis and integration also will be key. What will change in the 21st century? For one thing, we are going to see a lot more emphasis on modeling and simulation, on trying to capture heuristics in modeling and simulation. Engineers also will need to learn more about nonlinear systems, because such systems represent the bulk of what we are dealing with. Object-oriented methodology, as opposed to object-oriented programming, also will become more important, allowing us to build libraries of objects that we can use to assemble systems. More and more, engineers will focus on reuse rather than on designing each system from scratch and making all the same mistakes along the way. I expect we will see systems that reuse perhaps 80 percent of their code from application to application. Engineers will need to tailor the final 20 percent for specific applications. This will be a system-of-systems world, and so we are going to have to deal with the interconnection and interoperability of heterogeneous systems. That's not a skill that is very pervasive today, but it will be part of the bread-and-butter operations of the telecommunications industry in the future. With all of the emphasis on the importance of software in the network of the future, we are really going to have to learn something about software-system architecture as the first deliverable. Some business issues will be part of the profession, too. Engineers are going to have to help their clients make their business case. We do some of that today, but we are going to have to do a lot more of it. We are going to see a great many more multidisciplinary teams doing projects. We will see the emergence of the "solution architect" as the client's advocate. These folks will need to integrate aspects of many different disciplines. In general, we are going to see a requirement for broader knowledge. As more engineers begin to work with smaller companies, they are going to have to acquire a lot of expertise that in large companies they might not necessarily need. The largest network operators are going to be software companies. Hence, it will be vital to have management personnel who are knowledgeable in software and how it drives their business. We are going to see design performance and design efficiency in some cases traded for our ability to manage complexity. And, finally, I think that we all recognize that this profession requires lifelong learning. Let me close by making a few observations on some innovations yet to be made in this business. One is the so-called lights-out, or peopleless, network. A second is telephone dialtone joined by video dialtone, data dialtone, and ultimately knowledge dialtone. There will be concierge-like directory services that begin to meet our exact needs by asking us to address specific questions and permitting us to impose specific constraints on the responses to our directory queries. For example, I may need a pharmacy, be willing to drive no farther than 2 miles, and need something that's open all night. That's the kind of problem a concierge-like directory service will address. Finally, search engines that answer questions as opposed to simply retrieving documents are an innovation yet to be made. I think that we are all familiar with a very famous quote from Niels Bohr: "Prediction is very difficult, especially about the future." In honor of Morry Tanenbaum, I would like to make another observation: It certainly is easier to predict the future than it is to change the past! Thank you very much. About the Author:George H. Heilmeier, a member of the National Academy of Engineering, is chairman emeritus of Bellcore. This paper is based on a talk he gave 13 February at the NAE National Meeting symposium in Irvine, Calif., honoring the service of outgoing NAE Vice President Morris Tanenbaum.