MORE DD zdnet.com
Internet 2002 August 31, 1998
The Net: Not Just Data Anymore
By Tom Steinert-Threlkeld
With packet networks becoming multimedia
highways, public communications stand to be
transformed in short order
Four years is not a lot of time. Of course, in August 1994,
U.S. Vice President Al Gore was talking about an
information superhighway and many experts were figuring
that cable television networks would be its foundation. The
big news was that Prodigy Inc. was going to provide access
to its online content via the World Wide Web. Monica
Lewinsky was a total unknown.
Way back then, there were only 3,000 Web sites in the
world, 3.2 million host computers and 30 million Internet
users. Now, there are 2.5 million sites, 36.7 million hosts
and 134 million users.
In that time, the use of Net technologies has removed all
doubt as to what basic communications protocols will be
used to tie together the rapidly increasing multitudes of
computer systems and networks around the world. The
Internet protocols now are the lingua franca of
interconnected computing. The only question that remains:
Will Internet Protocol (IP) become the lingua franca of all
electronic communications?
"The time has come. In fact, it's long overdue,'' says
Leonard Kleinrock, the University of California at Los
Angeles professor who, in 1961, published the first paper
on the notion of sending information in packets of digital
data. The U.S. Department of Defense's Arpanet, which
went on to become the Internet, came to life in Kleinrock's
UCLA laboratory in 1969.
Kleinrock's comment came shortly after Sprint Corp.
announced it was switching its entire local and long-distance
network to the kind of packet technology that is behind the
Internet. And while Kleinrock may be evoking a bit of the
pride of authorship, almost any large communications
company worth its cross-connects is scrambling to convince
investors that it is on track to become a leader in
Internet-based communications in short order. How
otherwise to dispel, for instance, stockholder chagrin at the
seemingly lackluster merger of GTE Corp. with Bell
Atlantic Corp.?
Make it clear that one of the driving forces behind the $68
billion merger was the 1997 acquisition of BBN Corp. for
$600 million, which made GTE Internetworking a leading
player in the era of packet communications.
In fact, as August 1998 draws to a close, the idea that, in
the next four years, the Net will have established itself as
the primary means of exchanging not just numerical data,
text and Web graphics, but also sounds, sights and voice
conversations themselves, is no longer some technology
wonk's spirit-induced "vision.'' That public
communications in the year 2002 will be synonymous with
the Internet, at this juncture, seems more realistic than
strained.
Multimedia Growth
Driven by growth of the Internet, data traffic is growing 10
times as fast as voice traffic. Indeed, WorldCom Inc.'s
UUnet Technologies Inc. subsidiary sees no relenting in
1,000-percent-per-year growth in data traffic on its
network. This has led John Sidgmore, its chief executive
officer, to conclude that, in another six or seven years,
conventional voice traffic on traditional telephone networks
will account for a scant 1 percent of all communications.
That kind of growth is possible not just because the number
of worldwide Internet users keeps escalating, but because
uses for the Net are becoming more complex. In a word:
multimedia. Video and audio files occupy much more space
on a packet network than do text and numbers, the main
constituents of today's electronic mail. But five-minute
movies soon may become routine attachments to e-mail,
each eating up tens of millions of bytes of space.
"People need bandwidth and demand bandwidth and will be
willing to pay for bandwidth,'' says David Isenberg, the
AT&T researcher who authored "Rise of the Stupid
Network,'' a critique of telephone company efforts to
create so-called "intelligent'' networks.
"I don't see [the growth in data traffic] letting up,'' says
John Curran, chief technical officer at GTE
Internetworking . "I see it accelerating.''
While traditional telephone companies, such as Bell Atlantic
or SBC Communications Inc., seem to spend most of their
strategic capital on merging with other traditional telephone
companies, new clean-sheet communications companies
with tens of billions of dollars in available funding are
building high-capacity networks.
These networks will not set up and tear down exclusive
circuits to carry calls; instead, they will send all traffic as
packets of data, from voice to video, from spreadsheet to
text document. They will use IP as the basic form of
creating packets, even if some choose to convert IP packets
to fixed-size packets to take advantage of a more proven
packet-switching technology - such as Asynchronous
Transfer Mode (ATM) - in the interim for moving data in
public communications networks.
In what seems like a sudden strike to a series of local and
long-distance networks that have developed over the last
century to handle one type of data - voice - the Internet
now has become more than just a medium for allowing
easier, more graphical communication between individuals
and offices. It has become the driver for a sea change in
how communications networks themselves are structured.
Just as the long-playing record and the 45 RPM single were
replaced by the compact disc, digital technology now will
replace analog sound wave technology throughout telephone
networks, not just on long-haul lines. Digital technology
promises greater efficiency and lower cost.
Economics 101
In the end, the Internet - if it becomes the public Net - will
win on the basis of simple economics.
Sprint, for instance, figures it will save 70 percent over
circuit switching with its packet-switching Integrated
On-demand Network. This will allow it to radically reduce
its charges. Companies - ranging from IP network gear
giant Cisco Systems Inc. to Microsoft Corp., with its CTO,
Nathan Myhrvold - openly talk about how, in the
packet-switching era, voice com- munications will be
"free.'' The theory: The cost of moving a bit will drop so
low that companies can sell packages of data services - that
will include handling voice calls - for less than what
companies now pay for their voice services alone.
Packet switching is more efficient than circuit switching,
because packets can be used to fill up all available
communications capacity; circuits remain open for the
duration of telephone calls, regardless of whether anyone is
speaking or if any data is being sent.
But it's not just a matter of sharing available capacity that
gives packet switching a commanding leg up over circuit
switching. There's the matter of the switches themselves.
With circuit switching, the pace of technological change is
impressive. So called "stored program switches" - such as
the Lucent Technologies Inc. 5ESS or the Northern
Telecom Inc. DMS - double their performance for the same
cost every 80 months, says Peter J. Sevcik, an associate at
Northeast Consulting Resources Inc., located in Boston.
More advanced ATM switches double performance in half
that time.
IP routers, however, double performance every 20 months,
and frame relay switches double every 10 months. When
calculated in terms of the number of bits per second per
dollar that can be moved, packet switches now beat circuit
switches - and the gap will widen.
Then, there's the software that runs routers and other
packet network switches. Programs that run conventional
circuit switches have been developed largely in-house by
telecom switch vendors, with input from their service
provider customers. But, in the age of IP, the ability to
write applications, utilities and other programs for
communications networks is wide open. Anyone with a
better idea can produce a product to improve
communications performance.
"IP brings thousands of entrepreneurs supported by billions
of dollars in the marketplace to bear on the problems of
communications. It breaks it down, breaks communications
down a piece at a time,'' says James Crowe, CEO of Level
3 Communications Inc.
That should help packet-switched networking based on the
Internet protocols overcome its biggest technical obstacle:
the fact that circuit switching offers higher quality and
more reliable connections for time-sensitive
communications - particularly for voice calls. Telephone
companies know how to run systems that serve millions of
people and businesses simultaneously, with few failures,
says Bill O'Shea, data gear chief at Lucent.
"They know how to keep the system up,'' Kleinrock says.
"The data guys are just beginning to think of this now.''
That's why many of the IP networks may actually, for a
few years, run on ATM switches. With ATM switches,
carriers not only can guarantee quality, reliability and
security, but they also can keep track of who is carrying
packets for whom and settle up bills later.
"There are a lot of things that we haven't comprehended
yet,'' Bell Communications Research Chairman Emeritus
George Heilmeier says about how to make packet switching
work as well as circuit switching.
Settlement procedures and quality assurance are still to
come in IP-only networking. A new version of the basic
Internet protocols will add mechanisms that can guarantee
higher levels of service for traffic that requires it. In the
interim, companies such as Level 3 are working, with the
current version of IP, on protocols that get around the
problems. Indeed, the quality of a long-distance call on
Qwest Communications International Inc.'s long-distance
packet network is so high that Baby Bells such as Ameritech
Corp. are trying to resell it.
In the face of this shift from circuit economics to silicon
economics, there may be method in the Bell telephone
companies' urge to merge their local telephone assets, after
all. As this era of end-to-end packet networks emerges, the
one choke point remains the last mile. For more than a
decade, long-distance networks have been digital, sending
data along glass fibers as pulses of light. But local networks
still rely on sound waves to get to homes and offices on the
tips of their copper lines. As long as copper remains the
first and last connection to the Net, local phone companies
will control access to customers and will be hard to bypass,
even in the packet era.
At present, competing local phone companies have largely
focused on picking off high-margin communications in
metropolitan areas where there are high concentrations of
corporate customers. Then, it is economically feasible to
lay down fiber to reach them or, in some cases, provide
"wireless fiber,'' as WinStar Communications Inc. does in
New York and other major markets.
Planning For The Future
Local phone companies don't plan to stand still, however, as
wired and wireless alternatives pick off prime customers.
SBC, for instance, says it is planning for its local networks
to be based on IP packets with an ATM switching "fabric''
by the year 2002. In addition, it plans to focus on providing
data services to businesses between now and then. It has
spent $3 billion in the past four years getting ready and will
spend another $150 million to $200 million per year over
the next three years.
Still, a circuit-switching company such as SBC will be
hard-pressed to abandon services it provides on networks
that, because of depreciation schedules driven by monopoly
regulation, are not yet all paid for. Besides, they can
command premium prices with these services. Currently, a
T1 high-speed connection - which can transmit data at 1.5
megabits per second - can cost a business $300 or $400 per
month or more.
"Will there be a pell-mell rush for IP in the local net by
SBC? Probably not," says Probe Research Chief Operating
Officer Allan Tumolillo.
SBC Executive Vice President for Corporate Planning
Michael Turner says that 98 percent of the company's
revenue today comes from circuit-switched voice
communications, and he figures that even when data
becomes more than half of his company's traffic, it still will
account for less than half its revenue.
The revenue disparity could slow down carriers in
deploying high-speed digital access technologies. Why
provide 1.5-Mbps access through Asymmetric Digital
Subscriber Line technology at $90 per month if businesses
start dropping T1 lines they'd been paying three or four
times as much for?
Or, why push virtual private networks if the end result is
that customers must abandon better-paying dedicated
circuits for cheaper IP-based temporary circuits? Only the
threat of losing the customer's business altogether.
Indeed, as the year 2002 draws near, the idea of end-to-end
packet networks that don't involve local phone companies at
all will be imminent, if not already in place. Level 3 and
Qwest, both based in Denver, plan to have complete local
and long-distance, packet-based networks in operation
nationwide over the next three years. These networks can
be higher-quality than the public Internet because they are
under one company's control. And because they use
Internet technologies, they can communicate easily with the
public Internet.
And they may be ideally situated as well, in the effort to
find ways to bypass incumbent phone companies' local
phone networks. Denver just happens to be where the cable
television industry is based, providing easy, face-to-face
communications with the companies that have the
highest-capacity local communications networks already in
place.
Adding momentum to the move to packets will be an
answer to the last-mile quandary that actually involves
sending local calls nearly 1,000 miles into the sky and back
down again. Close to launch is the Internet-in-the-Sky
network of Bill Gates and cellular pioneer Craig McCaw.
And while Teledesic LLC supposedly is targeting rural
customers unserved by competing metropolitan carriers,
many communications experts expect the company's
business plans to avoid offering universal service for all
kinds of packet communications, including voice.
"That makes the magic date 2002. That's when the wireless
alternatives enter the fray,'' says Larry Vanston, president
of Technology Futures Inc., an Austin, Texas-based
forecasting and consulting company.
But by that supposedly magic date, if packet switching is
indeed going to supplant circuit switching and the Internet
is to be its embodiment, the technology must learn to deal
with hundreds of millions of long-distance phone calls and
billions of local phone calls every day. Plus, it will have to
move streams of audio, video and other types of data at
speeds that will tax routers moving trillions of bits of
information per second.
The idea that the Internet in the year 2002 will be the
successor to today's public communications network is no
slam dunk. "The Internet hasn't really dealt with that kind
of thing, yet,'' Tumolillo says.
But, if usage keeps accelerating, it will have to learn fast.
Sprint Corp. can be reached at www. sprint.com
GTE Internetworking can be reached at www.bbn.com
Lucent can be reached at www.lucent.com
Level 3 can be reached at www.level3.com
SBC can be reached at www.sbc.com
Qwest can be reached at www.qwest.net
thanks, to jacksoo, for locating this article and pointing me there.
r1
|
