SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Voice-on-the-net (VON), VoIP, Internet (IP) Telephony -- Ignore unavailable to you. Want to Upgrade?


To: Frank A. Coluccio who wrote (1170)8/25/1998 1:33:00 PM
From: Frank A. Coluccio  Read Replies (1) | Respond to of 3178
 
This is a bit OT from VoIP, but an interesting, if not generic, article written about ATM in backbone networks. It also touches on multiple other technologies.

Enjoy, Frank C.
========

"ATM Shines on the Spine"

August 25, 1998

PC Week via NewsEdge Corporation : ATM Technology
has not been the sweeping success that vendors initially
anticipated, but it hasn't been a total bust, either. While
failing to garner much acceptance at the desktop level,
ATM is being widely used for network backbones.

ATM operates at up to 622M bps and includes inherent
redundancy that can route data around trouble spots on
a network, so it's well-suited to backbone applications,
which must handle the traffic from all departments
connected to the LAN or WAN.

In November 1996, Dr. Pepper/7Up Inc. needed to
upgrade its backbone, a 10M- bps Ethernet line that
enables 750 users in two buildings to share information.
The organization evaluated three backbone options:
ATM, FDDI and Gigabit Ethernet.

Robin Hinson, a systems architect and planning
engineer at the company, in Plano, Texas, determined
that FDDI lacked a viable migration path and Gigabit
Ethernet was untested, so the company chose ATM.

By last summer, Dr. Pepper/7Up had examined ATM
equipment from vendors including Bay Networks Inc.,
Cisco Systems Inc. and Fore Systems Inc. and selected
Fore's products. "Fore offered a flexible set of products
that could be easily integrated at an attractive price,"
Hinson said.

In addition to upgrading the backbone, the company
decided to upgrade to switched 10M-bps/100M-bps
Ethernet connections running on each floor in the two
buildings to serve its desktop users. The firm completed
the network in March, and Hinson expects the new
network to meet the firm's communications needs for the
next five years.

The city of Hartford, Conn., was in a similar situation,
with fiber-optic connections forming a backbone
connecting seven agencies, such as public works and
the city library, both located within a few blocks of each
other. The city had been using T-1 connections to link
the sites, but last year started to experience sporadic
network problems as its 1,000 users exchanged complex
files.

"In a distributed environment, network performance can
be a killer," said Al Teixeira, user support manager for
the information services department for the city of
Hartford. "Because there is no consistency in the type
of files users send from place to place, a network has to
be able to handle huge fluctuations in traffic."

The city uses the I-LAN router from Olicom Inc. to
connect the agencies and decided to upgrade from T-1
to ATM connections. The process started at the
beginning of the year and has been completed at four of
the seven sites. Teixeira stated that the T-1 network will
continue to operate in parallel with the ATM links until
the transition is finished later this year.

ATM is also being used to form WAN backbones,
linking offices in different cities. Prudential Insurance
Co. of America, Inc. has 75,000 employees and operates
data centers in seven cities that house a series of
mainframes, midrange systems and PC servers
connected by 50 T-1 lines.

At the end of 1996, the company determined it could
lower its communications costs dramatically by making
the switch from T-1 to ATM because carriers price ATM
circuits lower than T-1 lines. The company examined
products from leading ATM WAN equipment suppliers,
including Cisco, Fore and Nortel Inc., and, like Dr.
Pepper/7Up, selected Fore's wares.

Prudential then had to lease T-3 lines from several
telephone companies. Bernie O'Neill, vice president of
distributed computing at Prudential, in Roseland, N.J.,
said, "We wanted to go with one supplier, but the recent
Internet growth left many of those companies in a bind.
They have been unable to build their networks fast
enough to meet increased demand."

Prudential brought up its first ATM connection early
last year and completed the transition as the year ended.
The results surpassed expectations. " During the past 18
months, we have had less than 15 minutes of network
downtime," O'Neill said. In addition, the company has
already recouped the $1.8 million invested in the new
WAN.

The company may expand its use of ATM technology.
Currently, it uses Fast Ethernet for local backbone
connections but will need to upgrade that infrastructure
later in the year. O'Neill said the organization plans to
compare ATM, Gigabit Ethernet and EtherChannel and
select one by the end of the year.

ATM lacks integration ability

Although desktops represent the most common type of
network connection, ATM has garnered only minimal
acceptance there. The city of Hartford looked at
deploying ATM at the desktop but decided against it.

"We have a mixed networking environment and would
need to run Ethernet and Token-Ring connections along
with ATM," Teixeira said. "ATM cannot be easily
integrated with other networking techniques."

Jack Armstrong, an analyst at Dataquest Inc., of San
Jose, Calif., said standards designed to let ATM work
with other networking options have been slow to take
hold.

Pricing has been another problem. Rapid reductions
have lowered switched 10M- bps Ethernet prices to less
than $100 per connection and 100M-bps Ethernet to
$200; ATM costs approximately $500 per connection.

Server networking is another area where ATM
acceptance has lagged. With the move to sophisticated,
high-bandwidth client/server applications, companies
have been on the lookout for faster connections. Fast
Ethernet has emerged as the big winner in this market
segment.

At the end of 1996, Sonnenschein, Nath & Rosenthal
needed to upgrade its network infrastructure. "Users
had started to transmit imaging and video files that were
causing network performance to suffer," said Bill Vance,
network manager at the law firm, in Chicago.

The company selected ATM equipment from Bay
Networks for its backbone connections but decided to
go with Fast Ethernet for its server links. " There were
not a lot of ATM adapters available when we made our
selection, " Vance said.

Suppliers have concentrated on Fast Ethernet adapters
because it is a more lucrative market. Winning the
desktop and server categories has meant that Ethernet
products generate more revenue than ATM wares.

The Dell'Oro Group, a Portola Valley, Calif., market
research company, found vendors sold $8.8 billion in
Ethernet equipment in 1997, compared with $614 million
in ATM devices.

However, ATM's economic picture is not completely
bleak: The Dell'Oro Group expects ATM revenue to rise
to $1.3 billion this year due to its firm hold in the
corporate backbone arena.

Paul Korzeniowski is a free-lance writer in Sudbury,
Mass., and specializes in networking issues. He can be
reached at paulkorzen@aol.com.

<<PC Week -- 08-24-98>>

[Copyright 1998, Ziff Wire]




To: Frank A. Coluccio who wrote (1170)8/25/1998 1:46:00 PM
From: Frank A. Coluccio  Respond to of 3178
 
Interoperability the rage at CT forum
August 25, 1998

ELECTRONIC ENGINEERING TIMES via NewsEdge
Corporation : When formed in 1995, the Enterprise
Computer Telephony Forum (ECTF) was given a charter
to create interoperability agreements that would serve as
road maps leading to an open-systems environment for
the entire Computer Telephony (CT) industry. The
release of the H.100 and H.110 Interoperability
Agreements provided a significant step toward this
objective. Basically, these agreements detailed
specifications for a compatible telecom bus that has
literally ended the " bus wars" that forced hardware
developers to choose between multiple, incompatible
standards.

Most of the components shipping for the existing buses
were designed to be used in PCs with ISA, EISA and/or
Microchannel form-factor boards. As the industry
moved to PCI, ISA slots began to disappear. It was clear
that the industry was going to have to re-engineer all
existing hardware to accommodate the new architecture,
and this provided the ideal juncture for generating an
open-systems environment.

The effort was started in early 1996 when Brough Turner,
senior vice president and chief technology officer at
Natural MicroSystems, a principal member of the ECTF,
submitted a proposal to the ECTF board that the Forum
define a new, single-bus structure that would be a
compatible superset of MVIP-90 (Multi-Vendor
Integration Protocol), H-MVIP and the SCbus. The ECTF
technical committee chairperson endorsed the proposal
that was forwarded to the Hardware Compatibility
Working Group to tackle the project.

Lou Francz, a senior systems engineer for Dialogic Corp.,
pointed out that computer-telephony-systems
developers also played an important role in stimulating
the work required to develop an open-systems
specification. At that time, voice- and call-treatment
systems were moving in opposite directions, and
customers were being forced to make choices that would
require substantial upgrades later. Francz remembers
customers specifically saying that they did not want to
be committed to any single supplier forever.

Several companies were asked to participate in the
development of a series of agreements that would offer
customers as many open levels as possible. According
to Francz, the development of the H series of
interoperability agreements was a pragmatic test of the
ECTF's charter. The companies involved were all very
competitive, and each one knew that the closer the spec
was written to their designs, the more compatible their
particular products would be out of the box. Aware that
this would defeat the goal of the project, the group
worked hard to ensure that the resulting spec was not
based on any one existing standard.

The new group was quickly populated by
representatives from all the major bus companies as well
as the three key large telephony-equipment
vendors-Nortel, AT &T and Siemens. Involvement of the
larger vendors paved the way for the development of an
interoperability agreement for backplane chassis in
addition to PCI. Eventually, two projects were
defined-H.100 for PCI and later, H.110 for CompactPCI.

As a true open-systems standard, H.100 is expected to
be very successful. With a capacity of up to 4,096
timeslots, it supports backward compatibility with all
existing standards. Thus, current products did not
become instantly obsolete when it was introduced. Even
some proprietary standards are supported.

An increased burden was placed on the chip designers,
however. Since the new standard was a superset of all
the existing standards, new chips had to support all of
the subset standards. Specifically, chip designers were
faced with more timeslot abstracts in the switching
technology. Because the spec had to support a range of
timeslot timing speeds, the spec treated timeslots as
N-dimensional bearer channels, allowing them to be
bundled to create higher-speed channels. This extended
the development time of new products somewhat, but
carried no performance penalties.

Francz said that the immediate impact of H.100 was to
provide drop-in replacement capability at the hardware
level. This makes it much easier for smaller companies to
enter the now-unified market. Smaller companies were
once barred from entering the market unless they had the
necessary "corporate mass" to develop products
compliant to each standard. Now, with a single
specification to grab onto, it is possible for a variety of
companies to provide hardware devices that can be used
in a system with devices from other vendors.

According to Jacob Cepelinski, engineering manager for
Mitel Corp., H.100's greatest value was as a
stepping-stone to H.110. Although PCI was certainly
becoming the interface standard for PCs, the
CompactPCI platform would become critical for
high-density telecommunications systems. The
CompactPCI chassis offers a more reliable and
expandable solution than the cable-based PCI solutions.
It seemed obvious to Cepelinski that as the volume of
these chassis grew, costs would continue to decline,
making them more attractive to smaller organizations as
well.

The CompactPCI and H.110 standards now enable
telecom-equipment manufacturers to develop open
hardware that can include components (boards and
subsystems) provided by OEM suppliers with greater
ease of integration. In Cepelinski's view, the extra cost of
adhering to the new standards will be more than
compensated for by reducing the number of components
to be developed and by accelerating time-to-market for
new systems.

Cepelinski also noted that in addition to the specification
of a common telecom data bus, H.100 and H.110 also
define an embedded message channel that can provide
for the control and maintenance of peripherals that are
not bridged to the PCI or cPCI bus. This will enable
systems to proliferate the H.110 bus cost-effectively to
low-end peripherals that require only simple serial-bus
control.

Although the hardware-compatibility issues have been
addressed by the two specs, Cepelinski feels that there is
still work to be done in the driver area. Integrators need
enough drivers to be able to configure cards with the
variety of operating systems they will encounter. Board
manufacturers will need to provide the drivers, and the
silicon manufacturers will need to support them.
Cepelinski believes that drivers will begin to emerge
soon. He pointed out that the industry is in the very
early stages of design and production of boards based
on the two standards, since H.100 was released in May
1997 and H.110 was released in March 1998.

There is in fact much more work to be done in software
development beyond the needed drivers. System-level
interoperability issues, such as low-level switching, need
to be addressed, as does the question of how different
vendors can support a uniform interface to the server.
The latter will be addressed by S.300, the
service-provider interface, which will be published by the
ECTF when it is completed.

With the announcement of H.100 and H.110, the ECTF
has provided a clear path for designing compatible
hardware for the entire range of telecom requirements. It
is the designed-in compatibility of these two agreements
that provides the required flexibility to help fuel the next
phase of growth for the CT industry. The main benefit of
both of these specs is delivered to systems integrators,
who can now concentrate on configuring the best
systems to meet users' needs without worrying about
compatibility issues.

Natural MicroSystems' Turner notes that these
compatible solutions will stimulate the market to grow at
an accelerated rate, as standard solutions begin to
penetrate the very large markets now served by
proprietary solutions. The current market for voicemail
and IVR is estimated at $7 billion.

Copyright - 1998 CMP Media Inc.

By William H. Matlack Jr., Marketing Consultant, ECTF,
Fremont, Calif.

<<ELECTRONIC ENGINEERING TIMES -- 08-24-98, p.