Rivulet CEO Ed Kennedy on fine-tuning IP networks

Earlier this week, former Tellabs executive Ed Kennedy was named chief executive officer of Rivulet Communications, a new equipment startup aimed at improving service quality in IP networks. Kennedy described Rivulet’s approach in this podcast. He went into further detail in the below interview.

On Rivulet’s hardware: We have devices that go in the edge of the network. They don’t have to be located with the router, though we also have relationships with Cisco and a lot of the large government integrators.. They could be located at the customer premises, the ingress of the network, different spots in the network--it all depends on what the topology is. As the traffic comes in, we go out to the network and discover it. The IP network is very random. You’re shooting packets in. Sometimes they fall on the floor. Sometimes they get stuck in a router queue, which adds jitter, delay and latency. In the top queue, we say, ‘Look, only these signals can get through,’--these applications or calls or video links or whatever. If the network gets congested, we say, ‘Look, that can’t get through right now because there’s no more bandwidth.’ Or you can put it on best effort. In the medical case, for example, these files are streaming video at 30 Mb/s on a hospital network, which has a tremendous amount of traffic. If all of a sudden you’re getting stepped on--you know, everyone has had that YouTube experience, where you’re watching video, and it stops or gets blurry. You can’t have that when you’re doing surgery. We prevent that, and in doing so, we can let that traffic go through like it’s the only thing on the network.

On MPLS: We’re fully compatible with MPLS. One of the guys here says we’re kind of ‘MPLS at the edge.’ That’s somewhat of a misnomer. MPLS sets up these flows, and it’s very difficult to do adds, moves and changes there. However, you don’t do a lot of adds, moves and changes in the core, and that’s where a lot of MPLS is deployed. Out toward the edge, it’s the Wild West, where you send it and pray. We’re taking that last link--and that could be from the core to the metro or out to the service edge--and we’re making that where we basically--we kind of ‘TDM the traffic,’ if you will. We synchronize it. When you send it, you’re guaranteed it’s going to go through with no delay. It has a time slot [in which] it sends it out. So you don’t get any of that jitter or dynamic latency or delay you normally get through routed networks.

On funding: We have incredible support from existing VCs and a lot of interest in additional investment. In fact, I was out yesterday, talking to various people. Our capability to raise money I think is pretty big. That’s not an issue. This is a pretty big play. The telco market, which is a lot of my heritage, there’s a lot of interest there, but there’s also tremendous interest in the government area because they’re first-adopters. They were first with IP, ATM, MPLS. They’re very progressive in that respect. So there’s lot of activity there and a lot of activity in some of these enterprise applications, so we’ve got a pretty wide berth of things we’re addressing and because of that, we get money from a lot of spaces. So money shouldn’t be a problem.

On partnering: We’d [sell] through [vendor] partners. We don’t have to work with a router vendor. Gone are the days when the big carriers would deal with startups. That’s why we’re kind of heading toward the small or medium sized [carriers]. Even Tellabs, when I left, was a billion dollars in revenue, and that was considered small.