Nortel's Mumford Savors Switches

SAN JOSE, Calif. -- Future telecom networks will see voice, video, and data come together over an optical switched core, according to Greg Mumford, CTO of Nortel Networks Corp. (NYSE/Toronto: NT) (see Nortel Bets on Mumford). Mumford spoke here at Opticon 2002 on Wednesday.

Contrary to comments made earlier this week by former UUNet chief scientist Michael O'Dell (see O'Dell: Voice and Data Don't Mix), Mumford says it's inevitable that data and voice are going to converge in carrier networks.

"You might say that culture won't let us bring voice and data together... but services will demand it," says Mumford. Business customers want multimedia content and mobile services, he says, and that's going to force carriers to continue to move from "costly separated networks to a shared network infrastructure."

One thing Mumford and O'Dell have in common is the belief in a technology that most in the industry think is in hibernation. Mumford, for example, sees business customers' demands for multimedia content and mobile services driving the adoption of optical switches by carriers.

His belief in optical switches runs counter to most industry opinion, which has it that the market for optical core technology has been put on hold indefinitely (see Nortel Shuts Optical Switch Effort, No Riches From Optical Switches , Cuts for Tellium?, and Core Optical Startups Chill Out ).

Ultimately, Mumford argues, all networks -- be they enterprise or carrier, wireline or wireless -- will need to run atop an optical core that supports voice, video, and data, consolidating network resources and saving costs all round.

Mumford says the migration to optical-switched cores is happening already, noting that Sprint Corp. (NYSE: FON) upgraded from a circuit-based to a packet-based network, replacing 132 existing central offices with 31 switching sites (see Sprint, Nortel Combine Sales Efforts).

Also, he observes, a financial company in New York went from duplicating dozens of databases in high-rent, hard-to-each locations to creating a large, offsite storage area network (SAN) that is accessible via high-bandwidth connections.

That application is significant, in Mumford's mind: "They're using the network to substitute for [computer] processing power." Once more enterprises adopt applications like SANs that call for big bandwidth, they'll start to move to carrier networks as well, he predicts.

Mumford envisions other aspects of these new networks, including optical Ethernet links to connect enterprise local area networks (LANs). These data-ready pipes will be run by software that can manage a network, saving the cost of sending technicians to do things like changing throughput rate on a connection, he says.

In the core, Mumford sees big optical switches with electronics for traffic aggregation, node consolidation, and service grooming. These switches will handle a broad mix of line systems from short reach to ultra-long reach, from 2.5-Gbit/s connections to giant 40-Gbit/s pipes.

Protocols such as Escon, Ethernet, and Sonet will all need to be supported in one switch, Mumford says. That's part of Nortel's pitch for its own Optera HDX switch, which sources say hasn't been gaining the ground Nortel had hoped it would (see Nortel's HDX is Here).

This part of Mumford's speech gave some attendees pause. "It sounds like they're adopting a 'God-box' approach to long-haul switching," says Paul Harrison, VP of marketing for Xtera Communications Inc., a long-haul DWDM equipment maker. Using one switch to handle transport and network services in the core narrows the field of vendors from which carriers can choose, he says. "I think service providers want more choice, not less."

Mumford, however, believes that as enterprise networks and carrier networks start looking more and more similar, the switches at the core must be able to handle anything. "If you need a switch for every different service, then you need a network for every different service, and you'll never get economies of scale that way."

Meanwhile, Mumford says, Nortel isn't advocating tearing apart existing networks to redo things. New technologies such as 40-Gbit/s long-haul links will need to be added piece by piece where they're economically viable. "I can't imagine somebody building a 40-gig overlay network..."

During a question-and-answer session after his keynote, Mumford acknowledged that Nortel has some work to do to address ongoing market needs. "We have the packetized portfolio now, we need to move it to an MPLS product set... We also need to get the HDX out there and demonstrate it, even in this time when no one's buying much optical."

After his keynote, Mumford told Light Reading that carrier spending is bound to pick up, but only after carriers improve their cash flow, something that, unfortunately, has stopped their buying patterns cold.

"They need to provide new services for new revenues," Mumford maintains. he feels that as new services are added, more equipment will be needed to support the evolving network traffic.

On that note, Mumford again bangs the drum for converged, efficient networks: "What are communications systems for if they don't bring people together and get them off the roads and out of the air?"

— Phil Harvey, Senior Editor, Light Reading
StartupGrunt 12/4/2012 | 9:53:40 PM
re: Nortel's Mumford Savors Switches How many millions of pitches have I heard that promised "voice data video" all in one switch!

This is not vision this is me too.
Belzebutt 12/4/2012 | 9:53:38 PM
re: Nortel's Mumford Savors Switches How dare you disrespect The Mumford!
edgecore 12/4/2012 | 9:53:30 PM
re: Nortel's Mumford Savors Switches Now I get it...if carriers could only:

-deploy SAN services
-deploy optical Ethernet services
-go from 132 CO's to 30

Nothing has changed in Nortel's message from 2 years ago!

Everything he said was mega obvious...and guess what...the world will not ever care to see 40Gbs until Cisco releases the HFR... The T640 shows promise to make 40 Gbps happen but, it does not have enough market share or push behind it to compete with the follow on to GSR (Not a CSCO fan but 70% vs 30% speaks volume in the core routing biz)....but maybe Nortel has other views..maybve the OPC will come back with a 4 in one 10 Gbps Ethernet combiner...:-)

BobbyMax 12/4/2012 | 9:53:29 PM
re: Nortel's Mumford Savors Switches The problems of convergence of Voice, data and multimedia is very old. To provide these services one certainly does not need optical network to provide and provision these services. In fact, convergent services over SONET/SDH than over optical networks.

The cost reduction is the main objective of all RBOCs.
Nomoredemo 12/4/2012 | 9:53:28 PM
re: Nortel's Mumford Savors Switches "It sounds like they're adopting a 'God-box' approach to long-haul switching," says Paul Harrison, VP of marketing for Xtera Communications Inc.,
Mumford has a reputation to deliver what he says . Its not always ontime but usually the nice charts turns into reality that usually sell. That being said im convince that he will deliver its vision which is not the case of all the players.
Prizm 12/4/2012 | 9:53:20 PM
re: Nortel's Mumford Savors Switches "...and guess what...the world will not ever care to see 40Gbs until Cisco releases the HFR..."

I don't follow your logic here. You seem to be saying that 40G transport gear won't take off until routers have 40G interfaces. The problem with this logic is that the highest rate that can be added/dropped from a 40G transport box is 10G. So, since today's routers have 10G interfaces, there should be a demand for 40G transport gear today!

Of course, it may be more economical to bypass the transport gear and directly interconnect routers with DWDM gear. But this is not what Mumford is talking about, as far as I can tell.

hitekeng 12/4/2012 | 9:52:52 PM
re: Nortel's Mumford Savors Switches Mumford hit the nail on the head and then hit his fingers when he says-->""They need to provide new services for new revenues," Mumford maintains. he feels that as new services are added, more equipment will be needed to support the evolving network traffic...."

In this current downturn, service providers are more than ever focused on offering new revenue-generating services YET at the lowest OPEX/CAPEX spending possible. They are in other words leveraging existing networks infrastructures to offer all such new revenue-generating services with limited hardware and software upgrades. Introducing new technologies or platforms in the RBOC world is a process of at least a year (if you are lucky), and even then, such will come at a cost.

Take for instance, the much hyped Optical Ethernet (OE) services that Mumford refers to in the convergence. Just to "converge" such a service onto a SONET MSPP platform will require millions of dollards to have it supported with OSMINE (i.e. need to migrate to Telcordia's expensive NCON...). So with cost tradeoffs of keeping OE over Ethernet switches versus consolidating over a MSPP and then optical switch, status quo will win for the next 2 to 3 years and this is where Mumford's and the industry problems are.

Sign In