So one of our biggest challenges is how do I free up resources in certain areas and move them to others? The speed of this change is dramatic. So new business models for pay-as-you-go, software-as-a-service, recurring revenue capability, deferred revenue. They all sound great, except I've got to pay for them today, and the profits are in years two, three and four. I have to free up resources and speed up decision cycles to move into these new markets.
Is this application-centric approach bigger than SDN? Does it relegate SDN to a technology that enables that?
We view software-defined networks as just one element. Think about the combination of SDN and NFV, which is service providers, in terms of network functions virtualization. Think about the cloud concepts, and then think about it as an architecture where you bring together servers and routing and switching and software and storage, first in the data center and then virtually throughout the whole cloud. It's all of these together that provide that - your ability to provision and bring services to market much quicker, in seconds or minutes as opposed to years. Without architecture, we think that doesn't work. How many CEOs do you think say: I want software-defined networks?
Not very many.
Or how many CIOs?
Not very many probably.
I'll tell you [what they want is] programmability. What they want is to be able to control operating costs and the CAPEX costs, the ability to be able to meet their business needs at a much faster pace. This is just one of the variables in that equation. We played defense for a year on SDN rather than just embrace it and say here's where it fits and here's where it doesn't, here's how it fits in the architecture. All of our ASICs we bring out are going to be SDN-compliant. Look at our next high-end ASICs which will generate the routing technology for the Internet of Everything, 4.1 billion transistors on this. [Ed. Note: Chambers shows ASICs.] It has 11 million lines of code. It cost $250 million plus to develop, half software, half hardware, the ability to take a chip architecture for a [Catalyst] 3850 but expand it to other products. We just announced it for the [Catalyst] 4500.
When you go to service providers, SDN does one thing for you. NFV does another set of functionality for you. And candidly, the cloud does another. What we're going to do is combine the cloud with the data center architecture, with SDN and NFV all together, and it will be a play at orchestration.
Let me ask you a couple questions about how you get to that architecture. One is that today you've got three data center platforms or fabric architectures, of Catalyst, Nexus and, now, Insieme. Are customers going to be confused about how they get to where you're going from these different platforms and offerings?
Any time you offer multiple approaches to a market you've got to spend a lot of time with your customers, understanding the pluses and the minuses for each way you pursue them. What Cisco has always done is make this transition pretty smoothly. The results speak for themselves in terms of our ability to do this. We moved from the GSR 12,000 [Ed. Note: a core router for service providers] to the CRS and now we're up to, what, 750 accounts on it, when people thought we might be into maybe 10 or 15.
Sign up for CIO Asia eNewsletters.