No, Basil Alwan, president of the Alcatel-Lucent (NYSE: ALU) IP division, didn't make our recent Top 10 Movers & Shakers list.

There's a reason -- he's had his moment, and you've heard enough about him already.

That's not stopping us from splashing a few more pages, and a picture, on him right here, though.



As CEO of startup TiMetra, acquired by Alcatel in 2003, Alwan and his crew built a router architecture that's managed to crash the party formerly dominated by Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (NYSE: JNPR), and it's now a pivotal element to AlcaLu's 100-Gbit/s plans. (See Alcatel & TiMetra Seal the Deal and Alcatel Router Revenues Surge.)

Alwan will also play a highly visible role in AlcaLu's converged packet/optical strategy, moving toward carriers' goals of getting the two types of networks to better cooperate. (See AlcaLu Makes Its Packet-Optical Move and Top 10 Movers & Shakers.)

On the router side of 100 Gbit/s, AlcaLu wasn't the first to announce, but it claims that the trials it's won have been substantial and have come at Cisco's expense. (See AlcaLu Trash-Talks Cisco on 100G.)

Alwan and Lindsay Newell, another TiMetrite who's now a vice president of marketing at AlcaLu, wanted some face time with Light Reading to state their case for declaring the 7750 Service Router a 100-Gbit/s leader, even though nobody's shipping 100-Gbit/s interfaces yet.



So, recently, Light Reading showed up at AlcaLu's digs, in a quiet corporate park in Mountain View, Calif., to discuss the platform. (Initially with Newell, and joined soon after by Alwan.)

— Craig Matsumoto, West Coast Editor, Light Reading

Light Reading: When it comes to 100-Gbit/s, do you feel underappreciated by the media in general...

Lindsay Newell: Yes.

Light Reading: ...or just us?

Newell: The media in general. I'd put some of it back on us. Alcatel-Lucent's a very broad company. We're on multiple fronts. Juniper is kind of a highly focused IP networking company, so you'll get their whole corporate PR on top of you for anything to do with routing. Cisco's still a router company in many regards. You look at what happened with the CRS-3 announcement. It was a new line card -- but they had the whole corporation pushing this line card.

[Basil Alwan enters, uneventfully.]

Newell: We still feel 100 Gbit/s is a potential discontinuity in the core that presents opportunity. Especially having 100-Gbit/s IP and optics together. That's something, at 10 Gbit/s and 40 Gbit/s, that nobody really had.

The fact that we can win a major Tier 1 carrier tells you that the [Cisco] ASR 9000 is not where it needs to be. The Verizon Communications Inc. (NYSE: VZ) SES [Switched Ethernet Services] deal is interesting because there are a lot of [Cisco] Catalyst 6500 metro Ethernet networks around the world that are eight to 10 years old now and ready for replacement. (See Verizon Tests 100G With AlcaLu.)

Alwan: It's always a hard thing to do when you have an existing product line, which is where Cisco was maybe two years ago. At some point, the heat got so high, because it was an aging platform, that they had to make a transition. And what makes it difficult is that for any new platform in this space, you also want to do a modern software architecture, so you can do features like high availability. It necessitates rewriting a lot of the code, and then all the features have to be put on that new platform.

I'm thinking in terms of an analogy of selling a car. You've got to put air-conditioning in it, and you've got to put all the features on it to sell it around the world. So the question is: You've now built this new next-generation foundation, but it's only useful to the extent that you build out all the features. The ASR is certainly not an uninteresting platform, but they over-pitched it. With Juniper, it's a similar thing with the MX960.

LR: When you start a company like TiMetra, how far out does the roadmap go?

Alwan: When we started the company in 2000, we had a 20-year roadmap that we laid out.

LR: Twenty?

[Newell cracks up.]

LR: So, what, you have the US winning the World Cup in there?

Alwan: [Pretending to write up a roadmap.] "Obama wins election," "major banking fraud..." No, the real detailed roadmap is two to three years, and software is on a different roadmap than hardware. Hardware roadmaps tend to be three to four years, because you're working that far in advance.

But you certainly don't have 10 years of roadmap. What we did right was that we saw the mega trend, if you want to call it that, in the industry: IP/MPLS becoming the common bearer for all traffic. Therefore, the IP/MPLS nodes happen to be a different class of product -- a service router, not a router -- able to support high availability, able to manage QoS, able to do a lot of heavy lifting.

Another mega trend we saw was Ethernet becoming a service provider interface. At the time, it was almost all Sonet/SDH interfaces. We were the first guys to really focus down on Ethernet and give it the capabilities to be a service interface.

Newell: That, and the crystal ball and the golden tablets.

Alwan: What's interesting is that, for the first time maybe in the industry, one company is leading in routing, optical, and switching in a fundamental speed change. We have the only single-wavelength 100-Gbit/s coherent system available. (See Analyst: AlcaLu's 100G Game-Changer.)

And this [the 7750] is the only single-flow routing platform available.

You've got this 100-Gbit/s stream coming in, and in the case of our competitors, you have a chip that splits it to two or four different subsystems [because Cisco and Juniper reportedly need multiple packet processors to handle a 100-Gbit/s feed], which then bring it back to the backplane. That's why I don't call it single flow. You have to break it into multiple flows.

The problem with that is that it's inefficient over time, because you have to try to balance traffic across these four. You can't just put every other packet in a round-robin, because flows have to stay on one of these lanes. You end up with these caching inefficiencies. You end up with more board space, more power. Every time you replicate the packet processing complex, all of the tables -- which is a lot of the logic on board -- and all the memories have to be replicated. LR: How far can you keep pushing the 7750? There must be some limit to what the slots can do.

Newell: We're not there yet.

Alwan: When you break it down to the limiting factor, there are two things. There's the backplane itself -- but people even change backplanes sometimes. Our backplane is the same backplane. It's had a lot of staying power. We did perceive the need to hit a lot of speed, to do a lot of ramping.

Newell: It was grossly overengineered for 2003.

Alwan: The other thing is cooling. Cooling is a very important piece of a platform like that. We put out new fans over time, to get more airflow into the platform.

LR: When are you going to do a multichassis router?

Alwan: We've focused more on chassis density, and now we have leading chassis density, which opens up a lot of opportunities to us.

Newell: The thing is -- you're asking the question everybody asks. And our response has pretty much been the same thing over the years.

It's kind of interesting that Cisco did this big thing about "changing the Internet forever" and the only thing they announced is a 140-Gbit/s linecard, although they talked about something like 3 Tbit/s. (See What's So Big About CRS-3?)

How many of their customers have more than four or eight in a multichassis configuration?

LR: Not many. None more than eight, as far as I know.

Newell: Yeah.

LR: But multichassis is cool. Come on.

Alwan [laughing]: It's like Dodge having the Viper: Why?

[Tape is unintelligible at this point as everybody tries to get a zinger in.]

LR: Admit it: You want to put 72 of these together. [A reference to the maximum configuration of Cisco's CRS-1.]

Alwan: For guys who think about density -- we can do really well with them. If someone's focused on multichassis, that can be an issue for us, and a core opportunity. But frankly, as Lindsay said, what's the percentage of guys who are really focused on that?

LR: You've managed to get the 7750 into a lot of different markets.

Alwan: The core of our business is still around really one thing: the evolution of IP/MPLS to be the bearer for all traffic. But we've expanded, vastly, our focus. Mobile backhaul, caching. We've moved into DPI. The 7750 is the EPC [evolved packet core, for LTE]. The 7750 is the GGSN.

We put a big push into mobile backhaul, but you know what we find out? A lot of our mobile backhaul customers are using the same network to backhaul business traffic. So, having one platform as opposed to having different platforms is very important. That trend has played to our advantage as well, and that's where Cisco has a challenge -- to rationalize all those platforms down to one or two.

You don't introduce a platform and then tell them a year later or five years later that you're going to move them over to a new platform. They don't take kindly to that. I've tried that before.

LR: Did it work?

Alwan: It's very hard. The problem is that our customers operationalize these things. They have this entire structure of people who learn these systems, and it's very hard for them to change that. IT is almost always the biggest issue for them. So, when you introduce a speed bump, you're impacting their business.

Newell: It's one of the reasons why we're seeing such success with our 7705 or 7210 -- the fact that they use the same code. The same engineers know how to configure and troubleshoot them. The same management system.

Alwan: Like Southwest Airlines. Southwest Airlines only flies 737s. They don't have crews where this crew can't fly that plane. There is an analogy there. Our customers are trying to simplify.

LR: And you don't feed them.

Alwan: Exactly. Or, we'll give peanuts out with your VPN.

Newell: I must confess -- I thought five or six years ago that I'd be bored in the IP business. Because we'd have done everything. But my brain hurts every day, learning all this wireless stuff, CDNs, DPI, policy management -- I mean, there's so much new stuff to learn, still.

It's also interesting, and you've talked about this before, Basil, that the global expertise in IP is highly, highly concentrated in this very small geographic area. Whereas, most wireless, optics, DSL -- it's spread...

LR: ...all over the place.

Newell: ...all over the world. IP is highly concentrated here in the Valley. The three successful router companies? They're all based here. Two, you could say, moderately unsuccessful ones in Tellabs Inc. (Nasdaq: TLAB; Frankfurt: BTLA) [Vivace] and Redback [owned by Ericsson AB (Nasdaq: ERIC)]. Huawei's now going to build something here.

Alwan: What we don't tell people is: Any routing guy that tries to leave the area? We take him out.

Newell: Silicon Valley is the Bermuda Triangle of IP. Once you come in, you can't go out. I mean, I came for two years. That was 1998.

Back to Introduction