- 18 Hot IT Certifications for 2014
- CIOs Opting for IT Contractors Over Hiring Full-Time Staff
- 12 Best Free iOS 7 Holiday Shopping Apps
- For CMOs Big Data Can Lead to Big Profits
Network World - Cisco's massive user show, Cisco Live, is going on this week in Las Vegas. Padmasree Warrior, Cisco's senior vice president of engineering and CTO sat down with Cisco Subnet editor and blogger Julie Bort. They discussed Cisco's progress since Warrior took the helm as CTO, the reorganization of Cisco's engineering groups, Catalyst vs. Nexus, the so-called Cisco "tax" (its prices vs. the competition), network virtualization, Cisco's toe dipping into the brave new world of open source and the future careers of the Cisco CCIE faithful. What follows is an edited transcript of the interview.
BACKGROUND: A Cisco engineer says changes at Cisco are good
What are two or three concrete successes that Cisco has had since you took the CTO helm?
One was coming up with a strategy for cloud when the industry was still discussing what "cloud" means. We started articulating at Cisco the notion about private clouds, drove the partnership with VMware and EMC, created this whole notion of converged infrastructure. We started on this journey about three years ago, saying it is network, compute, storage coming together, working with VMware and EMC on the Vblock and obviously lots of people from all three companies developing that. Second thing, I would say is really shifting the company more towards an architectural focus, from being just a pure product company.
What does an architectural focus mean?
If you look at data center, nobody really buys specific products anymore. They look at a data center architecture that includes different products and how they are put together. It's not just a switch for the data center, but how applications fit and how does a management stack work on that. In collaboration especially. Cisco had started off as a company that provided different video and voice applications separately. We are now bringing together those sets of products to be more connected and work together.
Tell me a little bit more about the reorganization that John Chambers touched upon in his keynote today. What is your role and how you think it is going to affect customers?
We're putting the teams together. That's a change we made recently. We used to have three separate groups. We had a voice group, a communications group, a team doing Quad which was separate from the team doing Webex. We've brought all that together under a single leadership. It's painful - not popular decisions internally. We've exited things like e-mail and focused on where we need to go.
When I joined Cisco, there was a council of nine people leading engineering. About three years ago or so, we went to a council of five and now the council is no longer there. Engineering is led by two of us, myself and Pankaj Patel [Senior Vice President, General Manager of the Service Provider Group]. We truly co-lead the engineering group. It's not two in a box. We have a team that focuses on all core switching and routing.
We brought all the ASICs together, where we used to have many different groups developing ASICs for separate platforms. There was a lot of duplication, there was very little sharing. It is now all consolidated under one person. We brought all operating systems under one person. Ben Fathi [senior vice president, Network Operating Systems Technology Group] now leads all the platform independent software development for IOS, Nexus OS and XR. Collaboration used to be under, I believe, three people and the collaboration council had six leaders. Basically it is now Barry O'Sullivan [senior vice president, Voice Technology Group] that owns collaboration. In data center, we brought in David Yen [formerly Juniper's leader of its Network's Fabric and Switching Business Group and lead creator of Qfabric].
How will you know if these kinds of changes are working?
The metric we are using is accountability - there is single accountability for leadership now - for the roadmap, for gross margin, for growth and market share. Doing things like that and assigning accountability to people is how we're going to know that there is a difference.
But those metrics had mostly been growing under the old system ...
Yes and no. It was unclear who owned those metrics. There was unclear accountability. When you have seven people leading collaboration, we didn't know if we were selling video more, voice more and what are the trade-offs. One of the big things we heard from customers is that we needed to make our products interoperate better and we're not quite there yet. We're working on a roadmap. Cius is an example of that. Cius has Webex on it, it has Quad on it, it has Telepresence on it. All of the Cisco collaboration solutions work on it, whereas previously that would never have happened.
Today's big news was a reinvestment in the Catalyst 6500. Sounds as if there is some kind of backpeddling going on around the two switches. How do the Catalyst and Nexus fit together?
They are positioned differently. If you look at 6500 it is a broad-based platform for Cisco. It goes all the way from the data center to the access and plays in many parts of the network. And the Nexus 7000 plays primarily in the data center core. Nexus OS is a much more modern operating system.
One of the things causing a stir lately is Cisco's idea of a single-vendor network. Customers don't want that - they want plug and play and they want best in class.
What customers really want is something that works and something that lasts.
Well, they also want some ability to negotiate ...
Yes, for sure. But that doesn't mean you compromise quality. I talk to a lot of customers and what I hear is that they want something that is reliable. They want something that is future proofed. Going back to the 6500, it's an installed base and they want the ability to protect that investment. There will be some customers that purely want something that's cheap and there will be an opportunity for some of our competitors to compete there.
The idea is not new. The separation of the control plane, the idea of programmability in the network, these things have been around for a long, long time. Programmability is important. OpenStack is the open source way of doing that and we are active in that. It is really providing network as a service. OpenStack has done the compute side and the storage side and network as a service is the next piece. But the idea there is that programmability and developing APIs for people to work within the network is going to be very important. I think it will take time to evolve.
Open source is another big wave that IT is dealing with. IOS is not open. Where does open source fit into what Cisco's doing?
Our strategy there is creating APIs. If you look at massively scaleable data centers where they're doing very complex Hadoop-type, horizontal [topologies] then they want to be able to program the network one way. If you look at infrastructure as a service, managing networks as a service, then their programmability means something completely different. So our approach is to say, for infrastructure as a service, here is a set of APIs that will support programmability.