Nortel girds for net hardware battle
By John Dix
Network World, 3/2/98
Start-ups don't usually pop out of the box with 3,500 employees and revenue in the
hundreds of millions, but that's essentially what Nortel's new Enterprise Data
Networks unit did.
Officially formed last month, the new business unit is charging after the likes of
Cisco Systems, Inc., 3Com Corp. and Bay Net-works, Inc. with bold plans for
gigabit campus backbone products and wide-area ATM gear designed to appeal to
users with aging T-1 and sagging router nets.
Given Nortel's heritage as a supplier of carrier equipment, the wide-area ATM
approach isn't much of a leap. But it comes as something of a surprise that Nortel
would turn its back on its ATM knowledge and investment when formulating plans
for the enterprise campus data market.
''It's too late for ATM in the campus,'' said Albert Delorenzi, vice president of
technology and business development for the new unit. "That market will level off
or die. When you look at industry investment in Gigabit Ethernet, you can only
conclude that ATM can't keep pace.''
The Enterprise Data Networks unit becomes the fifth line of business for Nortel. The others focus on
PBXs, wireless products, broadband equipment and carrier gear.
Delorenzi said Nortel started tracking data revenues two years ago and last year chalked up sales of $785
million with year-to-year growth of 24%. The bulk of the revenue came from sales of Passport, an ATM
switch Nortel was installing in carrier networks as an edge switch and at some large enterprise sites as a
backbone device.
Thinking corporations were tapping out their T-1 networks and believing it had a good story to tell, Nortel
decided to stop playing around the fringes and jump into the enterprise market. It combined the Passport
business with frame relay access gear that came in 1996 with the acquisition of Micom Communications
Corp., and formed the new enterprise unit Feb. 3.
The units initial mission will be to sell corporations on the merits of swapping out T-1 multiplexers for
higher capacity, more versatile Passport boxes, said David Ramos, vice president of global marketing and
communications. ''We just sold a 17-node Timeplex T-1 replacement net and will try to show other
customers how this is a no-brainer,'' he said.
The Passport can use private line, frame relay, ATM or any combination of these backbone facilities. And
it can support any mix of corporate traffic, including multimedia and IP. Some customers are even using
Passports to build integrated voice/ data T-3 backbones.
Because 45M bit/sec T-3 trunks cost about as much as four or five 1.54M bit/sec T-1 links, large
companies can justify building T-3 ATM networks that support legacy and routed IP traffic - typically
supported on separate T-1 links today - as well as corporate voice traffic.
''When data traffic is low, the network supports more voice,'' Delorenzi said. ''When data spikes up, you
have the option of increasing voice compression or spilling it back into the public network.'' The upshot:
vastly increased data capacity and voice rates of about 6 cents per minute compared to the 10 cents per
minute charged by carriers.
''Half of our T-3 backbone customers are supporting voice,'' Ramos said.
Gigabit backbones
While the Passport gives the new unit an interesting WAN pitch, WAN products by themselves do not an
enterprise story make. The missing piece for the Enterprise Data Networks unit is the LAN component,
and here is where Gigabit Ethernet enters the picture.
Recognizing that this young market is already crowded and that Nortel faces an uphill battle fighting for
mindshare, Delorenzi said the plan is to try to change the rules.
''Most campus LAN environments are stratified into three layers of switching,'' he said. ''You have
departmental, intermediate and backbone switches. Our plan is to collapse this into one.'' That will deliver
performance and cost benefits in lower port costs and savings in terms of administering one vs. three
layers of switches, Ramos said.
Although short on details, Delorenzi said the idea is to deliver a high-capacity Layer 3 Gigabit Ethernet
switch capable of supporting multiple protocols, which is different from the IP-only approach of many
other gigabit vendors. ''After all, users still have a lot of IPX and SNA traffic,'' Ramos said.
Nortel would first pitch the switch as a backbone device that would immediately do away with the need for
an intermediate layer of switching by providing direct support for workgroups. And from there, Nortel
would try to push down into workgroups, replacing departmental switches.
The justification for jettisoning the departmental devices is the now popular belief that the 80/20 traffic
rule has been inverted by the World Wide Web. Where once 80% of LAN traffic remained within a given
LAN domain, now only 20% of the traffic does. Eighty percent of LAN traffic is being routed to remote
Web servers, either somewhere else on campus or out on the Internet.
With all that traffic bound for remote locations, there is little to be gained by keeping switches at the
departmental level.
Although Delorenzi said Nortel has applied for patents, he wouldn't say where they fit into the gigabit
picture. He also noted that Nortel wouldn't rule out acquiring technology as required.
An equally compelling piece of Nortel's gigabit vision is the plan to develop stringent network controls.
''Today the guy in the corner office has the same network priority, same capabilities and the same
associated network costs as the guy in the mail room,'' Ramos said. ''Never mind that the CEO is trying to
download financials and the mail room guy is trying to send a picture to his girlfriend. That's untenable to
the CIO.''
While the industry has tried to address this problem with remedies such as the Resource Reservation
Protocol, Delorenzi said these efforts are too limited. The answer is to pace traffic at the desktop, he said.
Asked about the timing of the gigabit push, Delorenzi said: ''The strategy is to get into the market this
year.''
|
