Part III: Avoiding 40 Gbps pitfalls in 100 Gbps evolution

FierceTelecom: You mentioned that 40 Gbps went through four painful stages. What are the vendor and service provider community doing to avoid the mistakes made with 40 Gbps technology?

Griliches: I think there are a couple of things. The market is in a tough spot right now where there's X amount of component vendors. There are fewer component vendors and there's a lot of margin pressure on them and the systems vendors aren't making huge margins either. How do you get component vendors to keep developing 100 Gbps technology that's cost effective?

One of the things that is helping is the Optical Internetworking Forum (OIF) saying 'here's the type of modulation scheme we're all going to work with, but you can still add your own special sauce to it. If we all do this modulation scheme people can start making some hardware and integrated components based on that modulation scheme and start leveraging the volume of those components which will bring the price down.' Several vendors have rallied together and said, 'yeah, that would be really helpful and we'll use that technology.' That takes a little bit of the pressure off and hands a method to the component vendors to make this basic product for us and make it in volume and we'll buy it in volume.

Also, people haven't tried to make 100 Gbps full systems, but rather they have used one modulation scheme. That's not to say that the other modulation schemes won't be explored over the next couple of years. I fully expect that to happen.

I think also that everyone has realized that a lot of this can be done in ASIC design. This means you're not looking at putting together a whole bunch of 'tweaky' optical components, but you're basically doing it in electronics. That's enabled a whole bunch of people to jump in and say we can work with that. The hard stuff in the optics space has been moved to the electronics. Not that the optics stuff is easy, but it does open up the doors for people to jump in and start working with it. While there still will be a serious cost for anyone to invest in 100 Gbps if they're going at it from a component standpoint, but I think it will help bring it to the market faster.

There's plenty of demand for it. The cable operators have been screaming about it for a while. I think for the last three to four years, the content providers were the ones that were yelling for 100 Gbps and now they've got the Tier 1s right behind them saying 'yeah we need it too.' Now that they have got a little more collective strength behind them, there is a market.

Part III: Avoiding 40 Gbps pitfalls in 100 Gbps evolution