'The business model for IT networks is under threat'

Each week we ask chief technology officers (CTOs) and other high-profile technology decision-makers three questions.

Image caption Pradeep Sindhu says network capacity will need to double each year

This week it's Pradeep Sindhu, the co-founder and chief technical officer of computer networking firm Juniper Networks.

Founded in 1996, the company is based in Sunnyvale, California, and employs more than 8,700 staff in 47 countries, generating profits of $618m from an annual turnover of $4.1bn (£2.51bn).

What's your biggest technology problem right now?

The biggest technology problem facing our industry is the fact that the cost of running networks is outpacing the revenue that companies derive from them.

This is not a healthy state for the industry to be in.

Most users of networks and infrastructure don't touch the infrastructure directly - it's relegated to the basement, it's not sexy.

But all the sexy things depend on the compute storage and network infrastructure. So the need for service providers to be profitable is not met.

Most network infrastructure was designed many years ago. These old networks were built for just one application - for example, the voice network for telephones, the television network for moving pictures, and corporate networks to connect a company's headquarters to its branches.

But this tendency to break things up into many small networks is no longer viable.

Technology has advanced: new networks should be general-purpose networks that can do everything. You have to put the complexity into the software.

However, there are places in the network where there is definitely a concern that the explosion of data volumes could overwhelm these networks. The access part of the network, the bits closest to the consumer, is in the worst shape.

Capacity will need to double every year.

What's the next big tech thing in your industry?

The next big thing is what we call the new network. It's fundamentally designed to solve the main problem of the industry, and has three elements.

Firstly there is the cost side, the capex [capital expenditure] and opex [operating expenditure] of the network, compared to the revenue. If you look at the revenue, you have to ask why it is not growing, and one reason is that the rate of innovation in the networking industry is not fast enough.

So to address the capex issues, you have to design systems that are as general-purpose as possible. There won't be any need for specialised video or corporate routers. That way you take most of the complexity out of the network.

The opex, meanwhile, has been growing very, very fast, caused by the complexity of the old silo networks, one for each application. As you try to make them work with each other, the complexity goes through the roof, and you have to do a lot manually.

Secondly, automation: If you look at the history of technology, you will find that everything that can be automated will be automated. We will design networks in as simple a way as possible to make automation simple as well. That's the only way you can affect cost.

Thirdly, simplify the software, have an open software platform as a basis for your network. That's the next big thing, and to make it happen all network protocols have to be open and standardised.

We think that if no-one in the industry solves this problem, then somewhere between 2014 and 2015 there is a crossover point that breaks the business model.

With a new model, companies can avoid this cross-over point where costs outpace revenues and you have no business anymore.

What's the biggest technology mistake you've ever made - either at work or in your own life?

I thought hard about this and couldn't pick out one above others.

I made a lot of minor mistakes, but my biggest has been not to listen to my intuition about the industry going in a certain direction soon enough.

Many times in the past 20 years I had a gut feel about the industry wanting to go in a certain direction, and not listening to that has been my biggest mistake.

Around 2001-2002 Juniper should have gone into the ethernet market - ethernet as an encapsulation layer... I think we should have gone into these markets earlier than we did.

I saw it in early 2001. I saw it coming, but we didn't act on it.

At least when founding Juniper, I did follow my instinct.

Related Internet links

The BBC is not responsible for the content of external Internet sites