Network start-up king expands his realm

When it comes to building network infrastructure companies, Wu-Fu Chen has the golden touch. He's founded or co-founded companies such as Cascade Communications, Arris Networks and Shasta Networks and sold them to the likes of Cisco and Nortel for millions or even billions of dollars.

Now he's helping to run four network start-ups, largely involved with supplying carriers looking to offer next-generation services for enterprise network customers. He met recently with Network World Senior Editor Tim Greene to talk about Santera, a new company building high-density voice and data switches for carriers, and the future of public networks.

What's Santera's plan to shape the infrastructure for next-generation carriers?

Santera as a concept started as, 'Let's build a product using the latest technology that could reduce the cost and size [of local switches] and deliver both data and legacy voice service in a very cost-effective way.' That's what I call super-class switching, with a form factor that is only one-tenth of today's Class 5 voice switch that could deliver not only Class 5 switch functionality, but new data services.

How long will it take for networks based on such new equipment to take hold?

It's started already, but to get to 100% it will take a long time, 10 years or longer. The technology is ready, but the deployment takes a long time. Fortunately, there are many new carriers that are more aggressive than the older carriers. They could take technology into a trial much faster.

What benefit will customers reap from Santera super-class switches?

Our super-class switch can take in ATM traffic in very high density. A Class 5 switch couldn't do that. So super-class can ride the new trend in broadband access [by supporting high-bandwidth links like DSL as well as voice]. That [combination] is unique as well as less expensive.

So customers will see less-expensive services?

They should. It will be less expensive to a service provider. How a service provider is going to share those savings is something to ask a service provider.

What will people do with all the bandwidth that products like Santera's support?

It's not going to be used so much in the interaction between humans and computers. What's going to use up a lot of bandwidth is computer-to-computer communications [such as accessing and updating data storage, databases and applications remotely]. That activity could burn bandwidth as fast as you could think about it.

Also when video comes, the bandwidth consumption will rise - you're not talking even 10 times, you're talking about 1,000 times the bandwidth. But first bandwidth has to become affordable. Otherwise, people are going to say, 'I don't need it because it's not affordable.' When you can afford it, you will find out it's like a drug. Once you get it, you'll want more.

What is the key technology to make this happen?

Wave division multiplexing technology makes it so fiber can carry a lot of bandwidth. But there is still a lot of cost involved in taking that bandwidth and partitioning it into pieces to deliver to subscribers. That's why there are a lot of metro access companies trying to create a way to do that and reduce the cost.

Which carriers will lead the way?

The CLECs. When you used to think about a [competitive local exchange carrier] you'd think about a small service provider, and that's not really the case now. Even [the established local carriers] usually set up another unregulated company that is a CLEC. Almost 95% of these new technology companies are targeting the CLECs because they are more aggressive and could go to market and deploy new technology faster.

How many CLECs will survive?

CLECs are building up a certain value, but they may not be able to continue as independent entities, and will be gobbled up. AT&T grabbed cable companies to do local distribution, and they're looking at wireless because they don't want to pay access charges [to connect their long-distance network to local networks]. Conceivably, they could grab another CLEC to provide that access.

Does it make sense for enterprise customers to trust service providers to run their networks?

In the past, that was not the case. Service providers really were not prepared to run a data network. Now I think the model of outsourcing is definitely a good one. The enterprise really cannot afford to deal with the technology changes - they come so fast. I don't know how the enterprise can keep up when changes come every week or day.

Companies like yours start up, then suddenly they are part of a Cisco, Lucent or Nortel. Do you expect that to continue with your latest batch of start-ups, or is there room for companies like yours to stand alone?

Cisco, Lucent and Nortel will have to continue to buy new technologies. Once you buy new technology, two years later you need a new generation of technologies.

To build an independent company is possible. To do that the company has to go out big and create a huge market cap. You can use that market cap to acquire rather than become the acquiree, and assemble enough product mix to stand alone. As a start-up you cannot afford to build a line of products, but without a complete product set it's hard to sustain a business.

You've started up a lot of companies in a lot of areas. What else interests you now?

The next step is building better services for service providers to deliver to subscribers. That's why I started Shasta [which made a carrier provisioning platform for IP data services]. Santera is coming more from the voice area with lower-level data services.

What's next?

I'm pretty active in fiber optics. The other one is wireless, which is much stronger in Europe and Asia. The U.S. is behind. The thing that is driving wireless is Internet appliances. There are going to be all kinds of Internet appliances showing up demanding different types of bandwidth.