Must-read 5/98 tele.com. Packet-switched infrastructure switch [ASND references. Info on Cisco "New World Order" program to be unveiled at Supercomm]
teledotcom.com
Who Knows? Everyone knows that today's
circuit-switched public network will give way
to a packet-switched infrastructure. Right
now, that's about the only thing that
everyone knows.
By Dawn Bushaus
Dawn Bushaus is Internet editor for tele.com. She can be
reached over the Internet at dbushaus@mcgraw-hill.com.
These are exciting, exhilarating, exceptional, and excruciating
times for anyone in the public network business. The first three
exes--the good ones--stem from the mind-blowing
opportunity that providers now have to create the framework
for what will be the next century's public network architecture.
That last ex--the bad one--stems from that very same
opportunity.
The anxiety that's now building among public network
providers isn't being caused so much by fear of the future as
fear of the unknown. Companies like AT&T and GTE Corp.
have been getting ready to make the transition from a
circuit-switched to a packet-switched network architecture for a couple of years now. The rise of the Internet pretty much
ensured that the convergence of voice, data, and video onto
one network--the Internext--will become reality. In fact, it's
hard to find a provider that isn't developing strategies and
technologies for the Internext (see "This Revolution Will Be
Televised," April 1998).
But all that activity may be the biggest cause for concern right
now. The reason: The longer individual providers and groups
try to define the future on their own, the harder it will be to
bring all those definitions together into a single, interoperable
public network. Some early trouble signs already are popping
up:
* A fundamental disagreement on how to build intelligence into
the network is polarizing Internext developers into
diametrically opposed camps (see "Bellheads vs. 'Netheads").
Most established telecom providers want to drive intelligence
deep into the network; most advocates of IP-centric
architectures--including many (but not all) of the new breed of
native IP backbone developers--favor big, dumb backbones,
with intelligence distributed out to end devices. [Sounds like Gary's vision] That division
extends beyond the telecom world into application
development, where champions of the network computer
model like Sun Microsystems Inc. (Mountain View, Calif.)
and Oracle Corp. (Redwood Shores, Calif.) are lined up
squarely against Microsoft Corp. (Redmond, Wash.) and its
PC-centric legions.
* Equipment makers also are forming splinter groups to
determine the public network's future. Not surprisingly, the
main battle lines are forming between mainstay suppliers of
public network equipment and upstart networking vendors that
want to extend their reach from the enterprise business into the
public network arena. But even within the two main vendor
camps, there are major differences in technology and
approach.
The lack of a cohesive vision could spell trouble for the
industry in the long run, industry watchers warn. "There is the
potential for a series of disparate networks that don't
interoperate," says Daniel Taylor, senior analyst at Aberdeen
Group Inc. (Boston), a market researcher. If service providers
choose to implement different network topologies, they must
find a way to make them interwork, he adds.
While that's not an impossible feat at this early stage in the
Internext development, it's something that will get more and
more difficult as things move along.
Not just talk
The voice over IP frenzy has pushed the packet-switched
public network agenda forward faster than most providers
could have expected. But it's not the near-term promise of
cheap voice service that has venture capitalists competing to
dole out hordes of money to startups with nothing more to
show for themselves than a fancy PowerPoint presentation.
And it isn't cheap voice that has the world's incumbent public
network carriers sweating bullets. It's the long-term promise of
a more flexible, economical public network that can carry any
kind of IP traffic thrown its way.
Those kinds of networks are now being installed by new
public network service providers like ICG Communications
Inc. (Englewood, Colo.), IXC Communications (Austin,
Texas), Level 3 Communications Inc. (Omaha, Neb.), Qwest
Communications International Inc. (Denver), USA Global
Link Inc. (Fairfield, Iowa), and Williams Communications
Groups (Tulsa, Okla.). [Nice list of ASND customers] The new carriers aren't saddled with
legacy public network gear, an advantage that established
carriers are well aware of.
"IP is to communications what the PC was to computing--it's
that fundamental a shift," says Dan Schulman, president of
AT&T WorldNet Services (Bridgewater, N.J.). It's a shift that
the telecom establishment can't ignore. "AT&T needs to lead
the way," Schulman adds. "Sitting back is not an option.
You're going to see us be very aggressive."
That aggressiveness already is in evidence. AT&T's research
and development arm, AT&T Labs, has started evangelizing
its vision for the intelligent network of the future. In that vision,
providers will implement a new kind of middleware at the edge
of the network to act as a gateway between end-user devices
or applications and the network's access and transport layers.
AT&T has developed a prototype of its middleware
architecture--code-named GeoPlex--which it is now talking
up at industry trade shows and within service provider forums.
The company presented its GeoPlex vision to the academic
and research community at a conference on the Internet 2
project last month.
AT&T's smart network blueprint is a neat evolutionary fit for
today's public circuit-switched network. But even within
AT&T itself, debate has been raging over whether that plan is
best for the public network yet to come. Last summer,
then-AT&T scientist David Isenberg created a stir when he
authored "The Rise of the Stupid Network," a paper that was published on an internal AT&T Web site. In that paper,
Isenberg argued that carriers would be wise to invest in very fast but dumb networks and leave intelligence up to the end
points, under the user's control--in other words, the kind of
network that is the exact opposite of what AT&T is now
advancing with its GeoPlex middleware.
"The nice thing about the Internet is that it isn't controlled by
telcos," says Isenberg, who left AT&T in January to start his
own consultancy, Isen.com (Westfield, N.J.). "We've never
had anything grow as fast, be as accepted or useful, or allow
for as much innovation as the Internet has," Isenberg says. "If
you start putting too much complicated stuff in the network,
then it's possible to end up destroying exactly what makes it so
great."
Some next-generation service providers share Isenberg's view
that the network should be as uncomplicated as possible.
"Incumbent carriers are facing two terrifying events: the
liberalization of the telecommunications market, which is
leading to cost-based services, and a revolution in
technology," says C. Holland Taylor, chief executive officer at
USA Global Link, a company that got its start in the
international callback business and is now building out a $1.2
billion global IP telephony network using equipment from
3Com Corp. (Santa Clara, Calif.). "Carriers are in severe
danger, and they want to add intelligence to protect their
control over services." A fast network run by smart
applications and smart end-user devices is the better way,
Taylor asserts. "We'll have pipes capable of carrying
anything," he says. "They'll be colorless until they get color
from what end-users put in them."
Central Intelligence
The fast and dumb network model is exactly the opposite of
the computing paradigm championed by proponents of the
network computing model. AT&T's vision embraces that
model and calls for industry cooperation to implement
middleware that would bridge the public switched network
and the Internet, allowing service providers to create services
quickly and deliver them over multiple backbones. "Right now,
we're dealing in a two-layer environment--there are
applications and transport, but there is nothing in between,"
says George Vanecek, chief scientist for Internet platforms at
AT&T Labs. "That means we have to reinvent the wheel every
time we want to deliver a new service. We can't scale services
to global proportions that way."
AT&T began working on its GeoPlex architecture three years
ago, when the company realized that IP was going to be the
future of telecommunications, Vanecek says. Rather than rely
on vendors to tell the company what it was going to need,
AT&T opted to figure things out for itself. The result is a
middleware architecture that can scale to support hundreds of
millions of customers, Vanecek says.
GeoPlex can handle functions such as user registration, access
control, authentication, security (encryption and digital
certificate management), protocol mediation, and usage
recording for billing, customer care, and network management
purposes. Applications like directory services, e-mail, IP
telephony, and unified messaging are integrated into the
middleware, but other third-party applications, such as global
directory services based on the lightweight directory access
protocol (LDAP), can work with the platform, Vanecek says.
Because the middleware is capable of managing and
manipulating protocols, it doesn't matter what kind of device a
subscriber uses, nor does it matter what kind of transport
protocols a service provider chooses to implement in the core
of the network. "Underneath, the service provider can be
running Sonet [Synchronous Optical Network] or ATM
[asynchronous transfer mode]," Vanecek says. "It doesn't
matter."
Contrary to what critics say, AT&T's proposed architecture
would make the public network more open, Vanecek insists.
"It's paramount that we open the telephone infrastructure in the
same way that the Internet is open," he says. Today, there isn't
a layer in the public switched network that allows software
developers to tap into a switch through open application
program interfaces (APIs), but middleware could make that
possible.
Besides articulating its vision at trade shows, AT&T is trying
to get other large service providers to buy into its idea by
working with the Multimedia Services Affiliate Forum
(MSAF), a group of global service providers and vendors
working to ensure that services interoperate across multiple
service provider backbones. Some of the world's biggest
telecom companies are MSAF members, including AT&T,
BT, Deutsche Telekom AG, and Nippon Telegraph and
Telephone Corp.
Middle March
Some central-office switch vendors now think that the CO
switch, which sits at the edge of the network acting as an
on-ramp, may be the logical platform for AT&T's middleware
concept. "We can envision a future where the central office is
a platform acting as a gateway for applications riding on top of
it," says Mark Wilson, director of strategic planning at
Ericsson Inc. (Research Triangle Park, N.C.). "That's one
plausible scenario."
A second option--which is more likely, at least in the near
term--is for Class 5 voice switches to become big voice
servers in a central office or point of presence (POP)
dominated by data networking equipment including
multiservice access concentrators, packet and cell switches,
routers, and voice, fax, and video gateways.
That's a scenario that Ericsson and many of its competitors are
taking very seriously, and it's a vision that's already on its way
to becoming reality as large service providers begin to deploy
multiple gateways in their POPs that will let them deliver
services such as click-through applications for Web-based call
centers and fax and voice over IP (VoIP)."The central office
of the future is going to have to provide standard voice
connectivity for the public switched network, which means
Class 5 switch functionality, but it also must handle fax and
data," says Chris Brickler, director of enhanced IP services at
GTE Corp. "It must also scale and provide a high level of
security like today's central-office switches."
GTE already is starting to make changes in the CO. The
company announced in March that it is deploying gateways
from NetCentric Corp. (Cambridge, Mass.) to deliver a fax
over IP service and a platform from PulsePoint
Communications (Carpinteria, Calif.) to deliver unified
messaging. GTE also plans within the next month or so to
begin testing a VoIP application called CyberID, which will let
computer users know who is trying to call them while they're
online. At press time, the company hadn't announced which
supplier it will work with for the CyberID application.
The state of transition
Other carriers also are beginning the transition. MCI
Communications Corp. is deploying gateways developed by
NetSpeak Corp. (Boca Raton, Fla.) to deliver a service that
lets end-users click on a button on a Web site to be linked to
a customer service representative. US West !nterprise
Networking Services (Denver) is testing a fax over IP service
and is working on development of a directory-enabled
network. And AT&T, Deutsche Telekom, Telecom Finland
Ltd. (Helsinki), Telenor A/S (Oslo, Norway), and Telia AB
(Stockholm, Sweden) all are either already offering VoIP or
are gearing up for trials set to begin this summer.
Three groups of vendors are trying to sell transition gear to
bridge today's public switched network and the Internet. The
first group is the big central-office switch vendors, including
Alcatel N.V., Ericsson, Lucent Technologies Inc., Northern
Telecom Inc., and Siemens Telecom Networks (Boca Raton,
Fla.). The second big group includes data networking
equipment vendors like Ascend Communications Inc.
(Alameda, Calif.), Bay Networks Inc. (Santa Clara, Calif.),
Cisco Systems Inc. (San Jose, Calif.), and 3Com. Then there
are the developers of VoIP gateways, companies like ACT
Networks Inc. (Camarillo, Calif.), Clarent Corp. (Redwood
City, Calif.), Hypercom Inc. (Phoenix), Mockingbird
Networks (Cupertino, Calif.), NetSpeak, Vienna Systems
Corp. (Kanata, Ontario), and VocalTec Communications Ltd.
(Herzliya, Israel).
The big CO switch vendors have a lot of experience in selling
to telephone companies, and they've proven they can build
platforms that are fault-tolerant. On the downside, though,
they don't have experience with IP. The router vendor crowd
also has some experience selling to service providers but,
perhaps more importantly, has a lot of experience with IP. A
negative for the router vendors is that they know little about
voice services, and their equipment, while highly available, isn't
as reliable as a CO switch. The smaller gateway providers
probably face the biggest challenge in that they don't have full
networking product lines to pitch. But they do have one very
important thing going for them: They have equipment in the
field. Many of the early Internet telephony service providers
are using gateways from companies like Clarent, Vienna
Systems, and VocalTec to offer VoIP and voice over the 'Net
(VON) services.
As these groups jockey for market position, partnerships and
acquisitions are likely. In fact, they're already starting to
happen. On the partnership front, Siemens Telecom Networks
has repurposed its EWSD Class 5 switch by integrating
3Com's Total Control Internet Gateway and making switching
components modular.
"We've broken the switch down into a Chinese menu," says
Pedro Calaco, senior product manager for IP telephony at
Siemens. "You can choose the components you want. It can
be a switch or just a server." Siemens has begun to
demonstrate its new EWSD InterXpress solution and will
begin trials with an unnamed telephone company in Europe
next month. The product is scheduled to be commercially
available this fall, Calaco says.
Taylor of USA Global Link says that provider is considering
the Siemens platform for its network because of the 3Com
connection. "We need a switch today, but it becomes a big
hunk of metal once it's viable to deliver voice over IP
end-to-end," Taylor says. "We don't want a stranded
investment."
Like Siemens, Lucent and Nortel are promising to turn their
Class 5 switches into voice servers and add data networking
functionality. Rather than forge a partnership to get the data
networking expertise they need, however, Lucent and Nortel
are buying those smarts. Lucent acquired Livingston
Enterprises Inc. (Pleasonton, Calif.) last year, and Nortel
announced in March that it is buying Aptis Communications
Inc. (Chelmsford, Mass.). As a result of the Livingston
purchase, Lucent rolled out a carrier-class access
concentrator in March that will be equipped with a VoIP
interface early next year. Aptis is scheduled to deliver a VoIP
interface later this year.
Among the CO switch vendors, Alcatel by far has the most powerful data networking ally: Cisco. The two suppliers
announced their partnership a year ago. So far, that
partnership has involved joint sales to service provider
accounts, with each of the vendors providing its own products
as part of the deal, says Larry Lang, vice president of service
provider marketing at Cisco. But the companies also are
working on some joint products, he says. Cisco also has been
making acquisitions that help it address the carrier market.
Last December, Cisco bought Lightspeed International Inc.
(Sterling, Va.), a move that has allowed Cisco to incorporate
Signaling System 7 (SS7) functionality in its gear.
Cisco is going to be a force to be reckoned with for
companies like Ericsson, Lucent, Nortel, and Siemens,
particularly as it begins to step up its push into the carrier
space. Cisco plans to use next month's SuperComm '98
gathering in Atlanta to unveil a new service provider marketing
initiative called New World Order. [Yikes...] Cisco isn't discussing the
details of the initiative yet.
"We're about to reach the crossover point where data
becomes the dominant traffic type," Lang says. "When you get
to that point, it's time to look at a packet infrastructure and
think about how to put voice on top of it."
Service agreements
In this brave new IP world, it's not just vendors that are
making deals and buying their way into new markets. At least
one of the new breed of public network providers has opted
to buy the gateway technology it's going to need to bridge the
gap between the public switched network and IP. Level 3, a
facilities-based service provider started by former MFS
Communications Co. Inc. chairman James Crowe and other
MFS veterans, announced last month that it is acquiring
XComTechnologies Inc. (Cambridge, Mass.), a competitive
local exchange carrier (CLEC) and software developer with
an SS7 VoIP gateway.
Not only is Level 3 getting the gateway technology it needs,
but it also is getting CLEC status, which is necessary if a new
service provider hopes to be allowed to interconnect with the
SS7 network, says Ron Vidal, senior vice president of new
ventures at Level 3. "All the talk these days is about peering
agreements in the middle of the network," Vidal says. "No one
is worrying about peering at the edge of the network, but that's
the on-ramp for the public switched network."
Level 3 is constructing a nationwide fiber network completely
free of circuit switches. The network will be built in phases
over the next three years. In the meantime, the company inked
a deal with Frontier Corp. (Rochester, N.Y.) to lease some of
the fiber capacity that Frontier has purchased from Qwest.
That deal will allow Level 3 to begin offering services such as
private line data, Web hosting, and colocation before the end
of the year. Level 3 plans to begin offering voice over IP
service next year, Vidal says.
With the transition to an IP-based public network just getting
started, there are many more questions than answers about
how to phase out the circuit-switched network it's taken 100
years to build (see "No Wrong Answers--Yet"). It
undoubtedly won't take 100 years to dismantle that network,
but nearly everyone agrees it probably will take at least a
decade for it to fade away.
Only one thing seems certain right now: The next five years will
be the most tumultuous the communications industry has ever
known. "Neither the Internet nor the public switched network
is going to look the same," says Aberdeen Group's Taylor. "I
don't know what they'll look like, but I know they won't look
the same."
All use of this service is subject to the Terms and Conditions of Use.
All Rights Reserved.
Copyright c 1998 tele.com, The McGraw-Hill
Companies, Inc.
All Rights Reserved.
Website designed by COMPUGRAPHIA
Home page designed by Dennis Ahlgrim.
Last Modified: 5-May-98
|
