Convergence Monster Mash [Discussion of LU and ERICY purchase of ASND]
techweek.com
With Cisco battling Lucent and AT&T-TCI facing off
against Sprint et al, small companies take cover
by Amara D. Angelica
The long-predicted convergence of datacom and telecom is here. Two formerly
incompatible technologies-the Internet for data and telephone-switching
systems for voice-are merging. The result will radically change business
communications, lowering wide-area-networking costs and creating new services
and career paths.
Cisco, Lucent, and Ascend have announced new products that bring these
technologies together and raise the bar in the highly competitive networking
business. In early June, Cisco announced four ATM switches for service
providers such as MCI, Sprint, and AT&T. Lucent and Ascend have also made
important new-product announcements over the past month.
Cisco, the world's largest data communications hardware company, has grown
into the "gorilla" of its marketplace, according to Silicon Valley consultant
Geoffrey Moore, co-author of The Gorilla Game (see "Cisco: the Making of a
Gorilla," Feb. 23 issue). A gorilla refers to a company whose technology is so
dominant that it defines its field-such as Microsoft in operating systems or Intel
in ICs.
But with its foray into ATM technology, Cisco seems to have grown from gorilla
into Godzilla, devouring companies left and right and challenging telecom giants
Lucent Technologies and Nortel. In fact, Cisco CEO John Chambers recently told
Computer Reseller News that it plans to add another 10 to 15 acquisitions to the
24 it has bitten off since 1993.
Several rapidly emerging technologies are driving the lucrative multibillion-dollar
enterprise ATM market: "voice over IP" (low-cost telephone calls via the Internet),
"video over IP," virtual private networks (see "The rush is on for Virtual Private
Networks," June 1 issue), and network commerce services.
To provide each of these services, service providers need to integrate
conventional IP (Internet protocol) technologies (such as routers, which are used
by both enterprises and Internet service providers to route Internet traffic) with
ATM (see "ATM vs. IP," page 19). ATM is already well on its way-a $686 million
market in FY '97, according to Portola-based market research firm Dell'Oro Group,
led by Fore, Cisco, Bay Networks, 3Com, and IBM, in descending order by
enterprise-switch revenues. The worldwide ATM market is projected at $8 billion
by 2002, according to San Jose-based Dataquest.
"ATM has turned the corner in terms of being accepted as a viable technology for
service integration in wide area networks," says Tim Smith, Dataquest principal
analyst, WAN. "One only needs to look at the recent Sprint and Bell Atlantic
announcements for evidence of ATM's growing acceptance with large service
providers."
Sprint plans to use ATM as a cornerstone of its new low-cost ION service (see
"Sprint Drops a Big Pin," page 1). Bell Atlantic plans to offer ATM services to all
of its business customers (see ba.com.
IP+ATM
But ATM by itself is limited because it can't efficiently handle IP (Internet) traffic.
This is a major priority for leaders in both the IP and ATM camps as they fight for
global datacom/telecom domination.
On the datacom networking side, San Jose-based Cisco ($6.4 billion 1997
revenues) is aggressively attacking the telephone-switch market, which is long the
exclusive province of telecom vendors. Cisco also claims it has developed a more
effective way to integrate IP with ATM.
On the telecom side, Canadian manufacturer Nortel has agreed to acquire Santa
Clara-based Bay Networks ($2.1 billion revenues in '97) to launch Nortel in the
router market-Cisco's bread and butter. Lucent Technologies is not far behind.
Preparing their own IP-ATM integration battle plans are Pittsburgh,
Pennsylvania-based Fore Systems; Newbridge Networks of Herndon, Va.; and two
other major Bay Area router vendors: Santa Clara-based 3Com ($3.2 billion in '97)
and Alameda-based Ascend ($1.1 billion in '97). Ascend is a rumored candidate for
acquisition by European telco-equipment giant Ericsson, which has recently
established a presence in North America.
According to Smith, Ericsson has not been a strong ATM player in the U.S., so
Ericsson would bring little value to Ascend's position here. "Ericsson has been
stronger outside the US, particularly in Western Europe, and this could be of
greater value to Ascend," Smith says. "The real impact of this acquisition would
be to create a large, experienced central-office provider with a strong ATM
portfolio to compete with Lucent, Cisco, Siemens, and Nortel for the service
provisioning business. Given recent industry events, Newbridge must be
considered as one of the next logical candidates for acquisition."In early June,
Cisco staked out new territory with its new line of "IP+ATM" products. These
enable service providers to connect enterprise customers' IP networks with
high-powered ATM networks. "These new products are fairly impressive," says
Smith. "Cisco has taken a price leadership position while offering increased
functionality. They're like the Chicago Bulls-until somebody shows they can
beat them, they have to be considered among the favorites."
These new products fall into two categories: "edge" switches, which connect to
enterprises or telephone-company central offices (CO); and "core" switches,
which connect edge switches to each other and to a backbone network. The
switches offer "tag switching" (an implementation of the emerging Multi-Protocol
Label Switching or MPLS standard), which are new techniques that allow IP
routing and forwarding on ATM-based networks.
The Cisco MGX 8800 edge switch handles DS-0 (64 Kbps) up to OC48c (2.5
Gbps) interfaces, with up to 45 Gbps of total switching capacity. It
supports a full range of service interfaces, including voice over IP, IP
routing and forwarding, SNA, and frame relay. Cisco claims DS-1 (1.5
Mbps) prices at one-half of the competition. Sprint and U.S. West are
planning to deploy the MGX 8800.
The BPX 8680 is a similar product but for very large service-provider sites,
supporting up to 16,000 DS-l lines.
The BPX 8650 core switch allows for immediate upgrade of Cisco's 8650
core switch for IP+ATM services.
The Cisco TGX 8750 core switch, due out next year, will handle up to
OC-48 speed and will scale up to terabit speeds.
On May 4, Cisco acquired Class Data Systems Ltd. (Israel, with a Cupertino office),
which has developed a "policy server" with integrated directory services. This hot
software allows any application to reserve ATM bandwidth without requiring net
managers to rewrite code.
The recent announcements by Cisco, Lucent, and Ascend "illustrate the tenacity
with which the largest players in the networking business are targeting each
other's traditional areas of strength," says Smith.
"The service provisioning marketplace will not be for the faint of heart. Cisco's
continuing development of larger, more functional multiservice platforms aimed at
large service providers directly challenges Lucent's dominance in the central
office. And Lucent's announcement of IP-based networking products directly
challenges Cisco's dominance in all things IP."
Cisco has another reason to worry: Lucent has filed a patent-infringement suit
against Cisco. The complaint, which "has no merit," according to Cisco CFO Larry
Carter, involves routers, frame relay network equipment, and ATM equipment.
That goes to the heart of Cisco's strategy of expanding further into the Internet
and telephone marketplaces.
Also, we all know what happened to Godzilla in the end.
Nortel-Bay Networks join-in prayer
Northern Telecom's $9.1 billion acquisition of Bay Networks is either a smart,
market-based transaction or a "Hail Mary" pass. Or both. It all depends on which
analyst you ask. But regardless of the reason, they agree the merger has
destabilized the industry's strategic balance and is certain to accelerate
tele/datacom mergermania.
Northern Telecom [Nortel], based in Mississauga, Ontario, Canada, and Bay
Networks of Santa Clara say their merger was based on the need for an
"end-to-end solution." Their joint announcement vaguely defined this solution as
"mission-critical Internet Protocol (IP) integrated networks that will reach anyone,
anytime, at any place in the world."
In reality, Nortel needs Bay's IP and packet-based technology to supplement its
circuit-switched systems, and Bay needs a larger partner now that it has clearly
lost the market-share battle with Cisco.
"It's a very smart move for Nortel, although I don't buy the hype behind their
reasoning," says Christine Heckart, vice president of Telechoice, a Tulsa-based
consulting and research firm.
Heckart says that Nortel, which has long battled with Lucent for the top position
in the older, circuit-switched telephone equipment business, has grown concerned
over moves by Cisco to branch out from its enterprise data communications
dominance into the telecommunications industry.
"If I am Nortel and see Cisco as a new competitor," Heckart explains, "then it
makes sense for me to buy the number-two enterprise WAN [wide area network]
provider. It gives me an instant footprint here and the opportunity to provide an
end-to-end solution."
Good old competition may have also figured into the equation, says Heckart. She
points out that currently, Lucent resells a significant amount of Bay Networks
equipment. "They [Nortel] could see the Bay purchase as a way to deprive Lucent
of products they need."
Heckart says an uneasy relationship could still exist with Lucent continuing to
resell Bay Networks equipment, but probably only if Bay remained as a brand. A
Nortel spokesman tells TechWeek that Bay would remain as a brand, at least for
the time being.
Regardless of whether or not Bay remains a brand, the merger can only whet
Lucent's appetite to buy its own end-to-end solution. Sources say Lucent has
been kicking the tires of both Ascend Communications and 3Com. But Lucent isn't
the only tire kicker in the arena. Ericsson and others have also been looking at
these and other datacom firms.
From its dominant datacom perch, Cisco sees the mergers as inevitable.
Don Listwin, executive vice president of the service-provider line of business at
Cisco Systems, tells TechWeek that: "As data, voice, and video begin to come
together, it's clear to the market that the Internet will drive this convergence.
Certainly, industry consolidation will occur, particularly as Old World companies
look for ways to compete in this Internet-driven economy. Cisco anticipated this
new competition and believes that our leadership in data networking positions us
well in this New World."
Cisco has no comment on the observations of Sand Hill Road insiders whose
less-charitable assessment characterizes the merger as a Hail Mary pass with a lot
of praying at both Bay and Nortel. These sources say they believe Bay is
desperate now that Cisco has pulled so far ahead of them. For its part, Nortel
realizes it has a solid position in circuit-switching equipment, which is rapidly
becoming obsolete.
"Their [Nortel's] biggest competitor isn't Lucent," says one investment banker.
"It's progress."
Look for more prayers and mergers over the coming weeks. And buzzwords.
-Lewis Perdue
More information on ATM
ATM For Dummies, Gadecki and Heckart, IDG Books, 1997: If "ATM" means
"automated teller machine" to you, this is the place to start. A bit out of date, but a
very readable, practical resource for managers and decision-makers.
IP Applications with ATM, Amoss and Minoli, McGraw Hill, 1998, plunges into
ATM internals. It lacks a management overview, but this is the most current and
complete technical treatise on ATM for networking professionals.
Demystifying ATM/ADSL, Busby, Wordware Publishing (due out this fall), offers
good coverage of ADSL and a thorough introduction to basic telecommunications
engineering, but it fails to explain ATM in depth.
ATM Forum (http://atmforum.com) is an organization dedicated to establishing
standards, educating, and promoting ATM, mainly to networking engineers.
Cisco ATM Solutions (http://www.cisco.com/warp/public/730) includes white
papers and other useful resources.
ATM vs IP
There are three big communications concerns in every enterprise: bandwidth,
access, and cost. Originally developed by Bell Labs and France Telecom, ATM
(asynchronous transfer mode) offers a solution for each of these concerns.
Previously, service providers had to use different networks for various information
types, such as voice, data, and video. ATM is now emerging as a more efficient
way for enterprises and service providers (national and regional telephone
companies as well as large Internet service providers) to efficiently send various
types of information over a single, high-capacity wide area network (WAN).
ATM software runs on "switches." These are special-purpose computers similar
to telephone switches (located in telephone company central offices). But ATM
switches are able to handle various types of information (not just voice) over
virtual connections to desired destinations-computers in other parts of the
world, in another building, or the road, etc.
ATM achieves its high efficiency by using a fixed-length 53-byte "cell" instead of
packets or frames of varying length, as in other networking solutions. The cell's
predictable length and destination eliminates wasted time required for hardware
and software to constantly check for packet or frame length and destination. ATM
is also asynchronous (thus the name), which means that various data streams
(data files, voice, etc.) share the transmission resource (fiber optic circuits, etc.) as
needed.
This contrasts with traditional time-division multiplexing (TDM), or "synchronous
transfer mode" (STM), which is used in most data-communication systems. STM
allocates a predefined slice of time to each information stream-regardless of
whether or not there's actual data to be sent at that moment. Not very efficient.
Another big plus with ATM is that it allows for establishing a "traffic contract"
that specifies a "quality of service" (QoS) level for each type of information. For
example, voice and video require a high-availability dedicated virtual channel to
avoid annoying dropouts and delays. But for a file transfer, an as-available QoS
channel is fine.
ATM was originally hyped by vendors as an end-to-end solution right down to
the desktop, but most experts now believe that ATM's overhead and complexity
limit it to WANs and large campus (multiple-building) networks transmitting a
variety of types of information.
The Internet works in a manner diametrically opposite to ATM. ATM virtual
circuits are "connection-oriented" (every cell in a session is transmitted over a
predefined, fixed route). TCP/IP, the Internet protocol, on the other hand, is
"connectionless"-routers forward data packets over various unpredictable paths
(depending on circuit loading) to the destination, where they are temporarily
stored and reassembled into the original information.
This process is very effective for routing around problem circuits, but it's often
painfully slow and inefficient, with unpredictable quality of service, as anyone
who has tried to talk over the Internet or listened to a RealAudio program quickly
discovers.
What about sending IP (Internet protocol) information over ATM? This has been
possible for some time now. But new techniques, such as Cisco's tag switching,
use instructions in the cell header that allow IP information to be efficiently sent
via ATM channels.
The combination of IP and ATM gives users the best of both worlds: compatibility
with the Internet while offering ATM's high-efficiency transmission and specified
quality of service for hot new services like voice over IP, video over IP, and virtual
private networks.
back to home page
Home | Article Archive | Job Search | Resources | Company | Subscriptions
c 1998, TechWeek and Metro States Media, Inc. |
