Britain switches its entire phone network to the Internet Protocol
The United Kingdom is certainly an informative society. There are placards everywhere saying that pubs are fully licensed to sell spirits, that technicians at computer repair shops are certified, that London’s West End theaters are registered charities. A notice at a bus stop reads, ”We’re sorry that the timetable you require is not displayed here at the present time.” On the London Underground, I recently heard this frightening but helpful announcement over the public-address system: “Be advised you cannot transfer to the Northern Line at Bank station, because of a passenger caught underneath a train. You would be advised to transfer to the Northern line here.”
The UK is certainly also a leading member of the information society, but for some years it was a laggard in broadband. As late as 1999, all but 1 percent of the nation’s homes accessed the Internet at dial-up speeds. “We were in danger of becoming the big, fat, dumb, cheap pipe company,” says Matt Bross, chief technical officer of BT Group, in London, which was formerly the national carrier British Telecom and is still the overwhelmingly dominant provider of the “pipes” through which the country’s phone calls and data flow.
By March 2006, though, 19 percent of UK residents were on broadband, a larger fraction than the United States can boast. Broadband is available to essentially everyone, something the United States and most other countries can’t claim. In September, high-speed access came to tiny Foula, population 40, the westernmost of the Shetland Islands and the last telephone exchange in Scotland to be connected to the Internet.
Now the UK is getting a complete telecommunications makeover, from Foula to Cornwall and from the network’s core to its edges, vaulting it ahead of every country on the planet, even data-obsessed South Korea. In November, BT began the Herculean task of replacing its existing telephone network with one based entirely on the Internet Protocol. When the £10 billion network project is completed, there will be no technical difference in the UK between the telephone system and the Internet, though they will be distinct networks for security, quality-of-service, and billing reasons. The BT initiative, which the company calls the 21st Century Network (or 21CN), will give the country a phone system that will be at once the simplest and most modern imaginable.
But that’s just the beginning. Under the hood of the network, so to speak—in its central offices and exchanges, its metropolitan aggregation points, and its core routers and switches—BT plans to shut down all 16 of its legacy networks by 2012, replacing them with a single network to serve the entire country, including its 22.5 million households. It will be a single global network, in fact, because BT has facilities in 170 countries. The 16 networks include X.25, which was the hot new thing in data-exchange protocols in the 1980s, and asynchronous transfer mode (ATM), the multimedia data protocol of choice in the 1990s, which, when used for digitized voice traffic, fashioned point-to-point virtual circuits—yet another capability the new network will have to re-create.
Although the upgrade will not create the fastest IP-based network around, it will be the most comprehensive one. No large incumbent carrier in the world—not in South Korea, not in the United States, not in Japan—has so much as a concrete plan for a complete conversion of its phone network to IP, even though all of them would surely acknowledge that such a transformation is inevitable.
Creating one network into which all the others have been melded is a monumental undertaking, because at BT, as elsewhere, the different networks have their own equipment, protocols, software, billing systems, and staff.
The reward for such a simplification is greater speed—speed in expanding the network and continuing to improve it, in repairing faults, and in training employees. It also promises savings. Since 2000, BT has been reducing the aggregate price of its services by an amount equal to £0.5 billion per year—almost US $1 billion at current exchange rates. According to Paul Reynolds, CEO of BT Wholesale, the unit of the company responsible for 21CN, the transformation has to recoup all that money for the company, as well as pay for itself.
No £10 billion corporate makeover is without risk. After extensive modeling, testing, and field trials, BT has started to unplug the traditional telco switches in the first areas to be served by the new IP equipment. In the event of an embarrassing failure, BT could revert to the old hardware, which remains in place. So far, the backup hasn’t been needed. On 28 November, the first customers on the new network made what to them were ordinary phone calls—mostly unaware of any changes to the system.
If you were going to roll out your 21st-century network in stages, where would you begin? How about a proud, industrious, well-educated region of the UK that has fallen on hard times? That would be South Wales. For a century it was the coal capital of Britain and, at least briefly, of the world. At the height of its productivity, in 1913, a quarter of a million coal workers produced so much coal that more than 10 million metric tons were exported, amounting to one-third of the world’s coal exports that year, according to one source.
The mines are closed now, and the scars of their heyday—the slag heaps and stripped hillsides—are almost entirely covered by newly planted fields of grass or forests of oak, birch, and hawthorn. But the hard-working, highly literate Welsh people remain, making South Wales a good test bed of homes, light industry, and government offices. The telecommunications needs of 350 000 households are handled through three supersize central-exchange offices in Newport, Swansea, and Cardiff, the capital of Wales.
The three superexchanges—dubbed ”metro nodes” in the new network—rule over six other exchanges. The nine of them, in turn, move phone calls and data in and out of 70 local central offices. Today, the copper lines between individual homes and those central offices carry two different services—traditional telephony and broadband in the form of digital subscriber lines, or DSLs. If a household uses both services, its copper line terminates in the central office at a splitter. Voice telephony goes through one side of the splitter to a mainframe-like computer known as a circuit switch—a bulky digital automaton that, in the 1970s, replaced the helpful telephone operators of old, who used to complete the circuit for a call by pushing a plug into a jack on a switchboard. Data traffic goes through the other side of the splitter to a device known as a DSL access multiplexer, or DSLAM.
Maintaining two side-by-side networks is wasteful—of money, technicians, and even space inside the cramped central offices. Back in the 1970s, circuit-switching equipment stretched from floor to ceiling and from one end of the exchange building to the other—a single long switchboard that formed a freestanding interior wall, with thousands of copper lines running into the front of the wall and thousands more running out the back. Fortunately, thanks to Moore’s Law, new equipment is always slimmer than the stuff it replaces. In 21CN, the circuit switch and the DSLAM functions are combined into a single device, a multiservices access node (MSAN). Throughout the network, but particularly in the MSAN, BT will employ a technique called multiprotocol label switching to simulate, within the IP network, the virtual circuits created by earlier digital technologies such as ATM [see diagram, "Slimming the Network”].
MSANs are the first sign of just how radical a change 21CN truly is. Although each one is relatively inexpensive (BT declined to give figures), there will be over 5500 of them by the time the network is completed. Because MSANs represent as much as 40 percent of 21CN’s total equipment costs, their provision was a coveted contract; only two were awarded. One went to Fujitsu, a Japanese manufacturer that has sold DSL gear to BT for some time. The second went to Huawei Technologies, a 20-year-old company that has seemingly come out of nowhere (actually Shenzhen, China) to win contract after contract in Europe, Africa, and elsewhere, in part by underbidding its competitors.
Stuart Hickin, one of the BT executives in charge of procurement for 21CN, tactfully acknowledges that having Huawei as a supplier “helps manage pricing and keeps everything competitive,” but he also says the Chinese company was thoroughly vetted. ”We looked at their working practices, their suppliers and subcontractors, their entire supply chain.” Huawei will also make some of the electronics for what is called “transmission”—equipment that turns the multiple wavelengths of light that flow out of fiber-optic lines into the ones and zeros of e-mail, Web pages, streaming media, and digital voice calls. The other transmission supplier is Ciena Corp., in Linthicum, Md.
The deal with Ciena, a networking specialist and leading manufacturer of optical-fiber equipment, surprised no one. Nor were 21CN’s other contracts—with Alcatel, Cisco, Juniper, and Siemens—astonishing. But the failure of Marconi Telecom, a key supplier to BT’s telephone network for 20 years, to receive a single 21CN contract was more than a disappointment. It pushed the company, the UK’s last indigenous maker of telecommunications equipment, into dissolution. In late 2005, Marconi’s equipment businesses were bought by Telefon LM Ericsson, the Stockholm-based telecommunications powerhouse.
Ericsson has long been a key vendor to BT, making its central-office switches and programming them with the software behind modern telephony services, including direct-dial international calling, call waiting, caller ID, voice mail, threeâ''way calling, call logs for billing, and emergency service. No other supplier could have been trusted to take the millions of tortuous lines of computer code Ericsson had developed over the years and program them into the gleaming new computers, called inodes, that will replace these old switches [see photo, "Dotting the ‘I’”]. They must reproduce all those services in the new network, so Ericsson was given an exclusive inode contract. It’s 21CN’s only single-vendor deal.
BT is wary of depending on just one company for any aspect of the network, in part for technical reasons. The core routing network, in particular, which sends trillions of bits of data every second into the transmission lines that run between metro nodes, is completely duplicated, with one set of terabit routers from Cisco and the other from Juniper. Although the spread of a virus into the core is very rare, once inside, the problem would shut the entire network down for days on end. But because the two router suppliers use different software and different security methods, as well as different hardware, it is essentially impossible for the same problem to infect both at once.
The British telecom landscape was greatly complicated—though greatly improved—in August 2005, when BT and the government regulators agreed to completely restructure the way phone service and broadband were sold to consumers. Some history is needed here. In the 1990s, to increase competition, the Office of Telecommunications mandated that BT distinguish different services—access to a consumer from the central office, long-haul trunk lines from one city to another, and so on—and sell them on a wholesale basis to other telecom providers.
This idea was tried in many countries at around the same time. It worked well in some places, such as France, somewhat well in the UK, and not at all in the United States. In 2003, oversight of telecommunications in the UK was combined with that of broadcast television and radio into a new regulatory body, the Office of Communications, or Ofcom. One of Ofcom’s first tasks was a strategic review of the relations between itself and, it said, ”the companies we regulate, and citizens and consumers.” Over the next two years, it hammered out with BT an accord dubbed ”The Undertakings,” a total of 230 separate agreements that took the general idea of unbundling and institutionalized it with a near-breakup of BT.
BT’s so-called last mile—the copper wire going from households to central offices—and most of the operation of the central offices themselves were split off into a BT subsidiary called Openreach. If it were a separate company, it would be one of the bigger enterprises in the UK, with 30 000 employees.
Openreach is now one of five subsidiaries within BT. The others are BT Retail, the unit that sells DSL to consumers; BT Wholesale, which sells high-bandwidth connectivity to large businesses and to other telecommunications providers; BT Exact, the company’s information technology arm; and BT Global Services, an enormous provider of telecom and other technology services to British and non-British multinational companies, making it a competitor of IBM, Electronic Data Systems, and other heavy hitters. It’s the Global Services subsidiary that gives BT a presence in so many other countries.
Openreach’s existence is supposed to level the playing field in the UK so thoroughly that if IP packets were round, they wouldn’t roll even a centimeter toward BT Retail or away from any other Internet service provider. If BT Retail signs up a new DSL customer, it has to arrange it with Openreach and BT Wholesale, paying approximately $18 per month—as does, say, Demon Internet, the UK’s oldest ISP and one of its largest.
Asked whether the playing field is fair, Clive Feather of Demon offers a cautious “It seems to be. It’s still getting a shakedown in areas like wholesale line rental,” which is an arrangement that lets resellers put all of a customer’s charges—telephony, broadband, mobile phone, and so on—on one bill. (Wholesale line rental is particularly important when customers want to switch from one phone provider to another.) Feather, whose business card has the logo of Demon’s parent company, Glasgow-based Thus, and in the place of a job title simply says “Internet expert,” notes that Openreach “was created to address a long-standing complaint that BT treated themselves better than others; it was BT’s last defense against an antimonopoly proceeding.”
If every part of BT is in flux, the one thing the company won’t be is a big, fat, dumb, cheap pipe company, to use Matt Bross’s phrase. What then will it become? If a metaphor for the company’s future is needed, one can be found close at hand: St. Paul’s Cathedral, its clock tower easily seen from Bross’s office. In London’s Great Fire of 1666, the nearly 600-year-old wooden-beamed St. Paul’s burned to the ground, its lead roof melting “in streams,” according to contemporary accounts.
Sir Christopher Wren, Europe’s greatest architect of the time, designed the resplendent edifice that stands there now, but he would not have been able to do so by modifying the existing church. The magnificence of the new St. Paul’s—with its oak arches, flying buttresses, symmetrical geometries, intricate friezes, and vast gilded dome—was made possible only by the destruction of the old one. So, too, with BT. If the company had tried adding piecemeal to its existing network, it likely would have produced a technical Frankenstein. Instead, the UK’s telephone network, like the great cathedral, is to be gloriously reborn.
To Probe Further
BT maintains Web pages on the 21st Century Network at http://www.btplc.com/21CN.
The ”Final statements on the Strategic Review of Telecommunications, and undertakings in lieu of a reference under the Enterprise Act 2002,” more commonly referred to as simply “The Undertakings,” is available at http://www.ofcom.org.uk/consult/condocs/statement_tsr.
BT has ongoing dialogues with other telecom companies, conducted under a program called “Consult21.” In May 2006, the business communications firm Light Reading looked at how Consult21 was coming along. Its report, ‘BT Servers Cough While Critics Squawk,” is at http://www.lightreading.com/document.asp?doc_id=95539.