[BITS] IP Telephony: Throwing out the telecoms rule book

Ian Andrew Bell (belli@bctel.ca)
Thu, 27 Aug 1998 10:26:07 -0700


Hey guys, here's an in-depth look at the challenges that face the new
AT&T/BT joint venture.

This new company plans to build a 200 Gbps IP network that will link 200
cities in 50 different countries. The network will feature an open
architecture that will enable customers to build custom networks and
monitor their virtual networks themselves. Neat!

"Added Nagel: "If you're planning a global scale network like this,
no
company, not even one as powerful as ours, can do it all. [AT&T and
BT]
have over 400 interconnections with other operators. We have to
develop
a complex brokering and mediation environment to enable us to
interconnect our networks. It's going to be huge task," he said."

What we're talking about is a global, flat, integrated network that will
guarantee service quality for the MANY different computer data,
fax-as-data, and voice-as-data routes. The REAL challenge is in
addressing each of the three (all of whom have vastly different
requirements on a network) using IP protocol sets in ways that allow the
fast freights through without substantially diminishing overal QoS. In
this way, IP Telephony network engineering has more to do with railroad
engineering than it does with Public-Switch engineering.

If you haven't already, there's a really good layman's approach to
IP-PSTN stuff originally called "The Rise of the Stupid Network" at:

<http://www.isen.com/papers/Dawnstupid.html>

-Ian.

===============================
Throwing out the telecoms rule book

CommunicationsWeek International, August 10, 1998

"We have an opportunity to do things differently," said David Nagel,
AT&T's
chief technology officer and the
company's "technology leader" behind the design of the AT&T-BT joint
venture's new network. "The customers are
clear about the kinds of innovations they want and we think we have a
good
chance to set standards [for the next
generation public network] and drive it forward)'

Doing things differently in this case means new technology, a new
corporate
culture, and a new set of business
relationships with customers, partners and suppliers, they claim. Out
goes
the circuit switching and conventional
interconnection arrangements with partners: In its place comes an
Internet
protocol packet network integrating
voice and data and delivered via software organized into new OSI-style
layers designed to sit "above" TCP/IP.

The companies claim this architecture will underpin radical changes in
the
way they can present services. But some
analysts worry about both the technical and organizational strains.

Perhaps the most radical aspect of the new vision is the proposal that
users be offered what amounts to a
constructor's kit option to build their own services by opening the
platform so that users can hook into application
program interfaces (APIs). "In the way in which we are driving this, we
are
arguably going to be more like a
computer company than a telco," said Len Tyson, head of network
engineering
within BT's global division.

"Essentially, an open computing platform is going to be part of the
network. This doesn't only enable us to create
innovative services-it enables customers and other service providers to
create their own."

And it will enable telecoms managers to directly monitor and manage
virtual
private services more directly. Tyson
said the sorts of functions he expected would have the most initial
appeal
to users would be tools that enabled them
to control security and changes to user addresses, passwords, and so on.

"For instance, we could have a way for customers to set quality of
service
parameters for users and applications:
They could also directly control security and make moves and changes,"
he
claimed, "The key thing from our side is
to make sure the OSS [operational support system] environment provides
visibility of user profiles."

Tyson said this level of user control had proved impossible to provide
under the old technology and distributor
arrangement. "In today's environment it's an administrative nightmare,
or
can be," he said.

The network itself is to be built on conventional IP, preferably fired
along by speedy new carrier-strength routers
when these become available. "The strategy certainly doesn't rest on
next-generation IF routers," said Nagel. "We'll
use whatever underlying network technology is available and appropriate,
including ATM."

And the architecture will eventually be adopted for the partners'
domestic
networks.

AT&T says much of the technology has been tried out as part of the U.S.
Internet 2 academic networking project,
that it has a real trial version up and running, and that it expects
vendors to produce standards-based "terabit"
routers- capable of throughputs of up to 1,000 gigabits-to support it in
about two years time,

"All the [multinational] corporates want IF virtual private networks,"
said
Nagel, explaining the choice of platform.
"And they want mail messaging and all the other enhanced services
available
through IF to run over them.
Unfortunately, at the moment, the user has to do the integration work to
develop applications like universal and
unified messaging."

But instead of the network operator controlling these layers to
prefabricate specific business. offerings-the
conventional approach-the companies claim their partners, other
carriers,
service providers and even customers,
will be able to dig into the underlying network to assemble their own
customized services by writing software to
hook APIs. These will offer access to functions such as email, Web
hosting,
voice and conferencing over IP and
Global Access.

To meet these needs, AT&T and BT say they are planning what amounts to a
global carrier-owned Internet, where
accepted open standards and protocols, such as H.323, which orchestrates
voice over IP, are controlled from four
new network layers to deliver business-strength services and
features-quality of service, security, charging
mechanisms, network management views and so on.

The last will see users able to log on from anywhere in the world to
make
or receive conventional calls, email and
other messages, such as voice mail. To meet these ambitious service
requirements, the companies will involve a
broad range of partners and suppliers.

"If you look at the size of the task, it's just not possible, not even
desirable, for one company to try and do
everything," said Nagel. "The Internet has proved the power of having as
open an environment as possible and
allowing everyone to participate."

"These proposals are actually very exciting and ambitious," said Graham
Finnie, an analyst at the Yankee Group
Europe, of Watford, England. "They are the first incumbent telcos to
move
so radically towards this idea of a unified
IP network, and it sends a very clear signal to the rest of the
industry.
Importantly, they're saying they're looking to
outsiders for virtually everything, which is a real change".

But Finnie warned that the plans were still essentially a
"technology-focused vision, which is not on its own going to
impress customers. The users, especially on the voice side, are still
skeptical about tinning all their services over
IP," even though "there is a strong desire to move to IP as a single
protocol" claimed Finnie.

"I'm sure it can be made to work and it does seem the logical way
ahead,"
said Tim Hills, an analyst at Cambridge.
England-based Analysts Ltd. "But it's going to take a lot of time and
effort-it's also going to be 'done on the fly,'
which adds further complications."

"As other operators all start to [move towards converged IP networks],
there's going to be pressure to have some
sort of standardization," he added.

Nagel does not underestimate the amount of work required to develop the
new
management layers to add the
operator strength to IP, but thinks it unlikely that an industry forum
will
emerge to structure the process. "The Nortels,
Ericssons and Lucents are already involved in the process of bringing
open
platforms and software to market," he
said. "And of course there are billions of dollars being invested in
start-ups in the United States...we plan to make
use of all this activity."

To develop the underpinning technologies, the new company will disperse
$1
billion per year from a special
investment fund. Tyson said this would be used to search out and fund
start-ups capable of plugging the mass of
technology holes still existing in the overall vision.

"The challenge we face is how to manage the process?" said Tyson. "How
do
you migrate from where we are to
where we want to be?"

Added Nagel: "If you're planning a global scale network like this, no
company, not even one as powerful as ours,
can do it all. [AT&T and BT] have over 400 interconnections with other
operators. We have to develop a complex
brokering and mediation environment to enable us to interconnect our
networks. It's going to be huge task," he said.

Finnie suggested that one potential threat to the alliance was that,
having
validated the IF approach, the company
faced "several years of just standing still," while some of its
competitors
get to market with similar services before
it. "This is actually a great opportunity for providers like Equant or
Level 3," said Finnie.

On the other hand, having a clear vision would give the new company
something to work towards, and increase the
chances of AT&T and BT maintaining the relationship in the longer term,
added Finnie.

AT&T-BT's new venture at a glance

* 50:50 joint venture combines their International networks

* $1 billion profit an $10 billion revenue projected for 2000

* Global voice and data based on existing Concert business plus new
products and services

* Global sales and services, Initially targeting the financial,
petroleum
and IT sectors

* International carriers' services business

* New global 200-Gbps IP network to 100 cities in 50 countries

* $3 billion assets--$1 billion from BT billion from AT&T

* 50:50 Investment totaling $1 billion n U.S. high-tech companies

IA Copyright (c) 1998 Information Access Company