As Nortel's new CTO, John Roese must think big. His task is to rally more than 12,000 engineers in product areas as diverse as Ethernet, optical and wireless mesh, and cellular infrastructure, and then outline R&D and technology strategies for each market - all while the company struggles to regain its footing.

For example, Nortel recently said it would cut about 1,900 employees in the next year in an effort to save more than US$1.5 billion by 2008.

As the former CTO of Cabletron and Enterasys, and most recently Broadcom, Roese has his name on more than a dozen network technology patents. He says that at Nortel he will be more of an advocate for the engineers and technology evangelist.

What is your role as CTO of Nortel?
Nortel has been without a full-time CTO for a while. Unfortunately, when you're in that state the technical community keeps churning out product, but it tends to not evolve that dramatically -- or the products evolve along different paths and you lose some of that synergy of being in a large corporation.

The things I'm being asked to do by Mike [Zafirovski, Nortel's CEO] are to work to unify the product line and communication strategy.

If you think about Nortel, they're one of the few companies that are in all the next-generation relevant spaces, both enterprise and carrier, wireless and wireline in both of those areas, and the unifying applications - whether it be voice or real-time communications in enterprise, or [IP Multimedia Subsystem] in the carrier side. That's a huge asset but only if you take advantage of it.

How do you get all of these different factions inside of Nortel to agree on one direction or strategy?
There are 12,000 engineers inside Nortel across all the different business units and projects. I hate to describe CTOs as cheerleaders but we really are - generally the role is to be the spiritual leader of the technical community.

It's the opportunity for the technologists to have someone speaking on their behalf - let's call it bragging on their behalf of all the great things the engineers are doing and giving them some visibility. Engineers don't like to be politicians. CTOs bridge that gap.

What are areas that could be better integrated, technology-wise?
The biggest is the linkage between the application frameworks that exist in Nortel and the infrastructure. We could look at it from either the enterprise or carrier side.

On the enterprise side, Nortel has a wonderful real-time communications offering, and it is investing really heavily in that area. They also have a large presence in the data communications, networking side. People may scratch their heads trying to figure out exactly why those two things work better together, as opposed to some other vendors' network with someone else's voice technology.

What are examples of that?
It's simple things like the discovery of devices automatically, automating the process of configuration, a consistent end-to-end security model. What I've found is that most of the pieces are there; there's no lack of core DNA. It's just the consciousness to put it all together.

That end-to-end experience, all the way from the bits on the wire through the user experience with the application, needs to be more advantageous when there's more Nortel content then when there's less Nortel content.

On the carrier side, it's the linkage again of the applications with the next-generation infrastructures. Carriers are a little more interesting because there is this tremendous evolution of new technologies.

WiMAX and mobile WiMAX are going to be very disruptive; but we don't want to make it disruptive to the end user. So therefore the frameworks on top of it, IMS, or an IP TV system or video distribution strategy - those technologies need to be able to span the test of time, as the underlying infrastructure changes through these fundamental evolutions.

How do your plans for corralling the technical teams at Nortel mesh with the R&D spending plan that was outlined before you came to the company?
I saw the spending plan in advance. I wouldn't have taken the job if I wasn't comfortable with the economics with respect to R&D.

The first thing is to focus on the relevant areas of communication going forward. I've been joking with a lot of people, saying there aren't a lot of token-ring networks out there and if I find any token ring inside of Nortel, we're going to find a way to not spend money on it pretty quick.

Now, that's an extreme case, but there's that gray area between what is relevant to the future and what needs to be migrated down.

Nortel is a big company and there's a lot of technology out there. The good news is that it's a big company, there's a lot of technology out there and there will be those categories.

What we're doing is looking at all the organisations and giving the business units - an opportunity to start with a clean sheet of paper... Because there has been a progression of events which do not necessarily result in optimal R&D activity - layoffs, restructurings... announcements that aren't good. Those things distract people from the day-to-day activities.

Some analysts say Nortel had, then lost, its position as the end-to-end alternative to Cisco in the enterprise. How do you get that back?
The market clearly wants an alternative to Cisco. The most important thing we need to do is to claim that birthright.

Nortel hasn't actually said it. When you look at the landscape out there, there is really exactly one company that has the scale and breadth and customer base, brand and technology to claim that position. Someone else would have to build it.

With Cabletron and Enterasys, I competed as a smaller company against Cisco. We did a good job carving out our niche - however, without the scale and brand and reach, it's harder to get to that next step of being the broader alternative.

Nortel has that already, or is as close as anyone could be right now. Step 1 is to make sure people understand how big Nortel is. How many enterprise customers understand that Nortel is a $10 billion company with 12,000 engineers and a 100-year history?

So we have to tell that story. The second piece is, if you have the scale, that's great, but now you have to go beat our friends at Cisco in the technology area. While they build acceptable technology, their products are not necessarily perfect.

Cisco has many of the same big-company issues that Nortel is facing. So we're kind of on an even footing to go at each other in the enterprise market.

What innovation expertise does Nortel have, unique to itself, that could beat Cisco? It's hard to say you're just going to do LANs and routing better than Cisco.
Right. No one's going to do Ethernet better than anyone else. That's not what it's about.

The things where Nortel has the advantage - if we choose to leverage it - is the fact that the company has the ability and understanding of building scaleable systems for carriers and enterprises that span the transport all the way to the applications. Cisco has been trying to put that together, but in areas such as carrier wireless or carrier application framework, that's a new territory for them. They're still figuring that out.

We have a better handle and experience on understanding the high-level unifying middleware and signalling technologies that are necessary. The term carrier-scale applies to enterprises in the way we build switches and routers and our security model.

Even if most networks don't increase the number of users on the network, in the next five to 10 years, the number of connected nodes will double or triple because everything is becoming IP-connected, and every user is using multiple methods and devices to communicate.

It's a more complex problem than just moving bits around, which heretofore has been Cisco's principal focus.

So what are the core technologies and non-core technologies?
There are five kinds of macro-level technologies we have to participate in, that we're already participating in.

On the enterprise side, it's enterprise wireless and wireline infrastructure. There are some great products, but we have to continue to innovate. On the carrier side, it's the same mix - wired and wireless. The last piece is the unified applications framework. This starts with middleware, which is unsexy but important; things like IMS are examples of that.

But also having a cohesive network management architecture is another example. Having a call processing and signalling layer. One of the most interesting things I've found in Nortel is a tremendous investment, heritage and expertise around SIP, which is the basis for IMS, but also a consistent signalling mechanism across an enterprise's real-time communication architecture.

As far as the things we don't need to be in, I have to go out find the specific cases of that. The things that are not part of that long-term vision, and that don't' fall into one of those categories. If we're still chasing them, and we're chasing legacy [technologies], we need to figure out a way to make that legacy bridge into these next-generation architectures as quickly as possible.

Because continuing to perpetuate many of those older technologies, without a clear bridge to the future, is actually a huge disservice to the customer.

When you talk about legacy, is TDM a technology that you might focus on migrating away from?
As long as you can migrate effectively, clearly the IP frameworks are much preferred over the long term.

Voice is an interesting example. The evolution of TDM to IP is a long process given the footprint that's out there. The user experience over a TDM environment, in a hybrid scenario, with IP backhaul, ought to continuously improve but to the point where there is no difference in the user experience, the security model, the reliability or continuity with the IP framework.

That results in the customer not seeing the transition to IP and just accepting it.

One of the things that went wrong with IP telephony in the early days was that the industry assumed that customers were willing to take these huge quantum leaps away from one user experience to a new user experience. . . . People don't buy a lot of traditional TDM-only voice systems these days.

At a minimum, they're looking at hybrid systems, and realistically they're starting to migrate and feel very comfortable that a bigger percentage of their voice infrastructure will be IP.

How are you working with other CTOs or the head technologists in the various business units at Nortel?
In the absence of someone in [the CTO] role, many of the business units, just through logical progression, basically had to have architecture support, and CTO-type people emerged in each of those areas. [Such as Phil Edholm, who is the CTO for the enterprise business unit.]

Clearly as we go forward, we want those people to do what they're doing. There will always be local architectures in the business units. But the centralisation of the vision and strategy needs to pull back into a more central organisation...

If the businesses are fragmented, and they're not interacting with one another in a formalised sense, then that's not consistent with the overall corporate strategy of actually getting leverage out of this very large organisation.

The good news is that everybody is very willing to do that. It was just the fact that there was no one there to unify that.