SAN DIEGO -- Cisco CEO John Chambers this week confirmed that the company is funding and plans to absorb Insieme, a start-up developing a software-defined networking (SDN) system.
Cisco has invested $100 million in Insieme, which is led by three Cisco engineers: Mario Mazzola, Luca Cafiero and Prem Jain. The three led two other Cisco spin-in start-ups - Andiamo Systems, which made storage-area network switches, and Nuova Systems, which developed Cisco's Nexus 5000 series data center switches.
Cisco will cap its investment in Insieme at $750 million, Chambers said. Cisco had similar arrangements with Andiamo and Nuova, which it acquired back into Cisco when the companies completed product development.
UP THE REVOLUTION: OpenFlow and SDN: Networking's Future?
"The team at Insieme is a team that's been remarkably successful for us," Chambers said during a roundtable discussion with reporters at the Cisco Partner Summit. "If you watch what they've done, they helped us do that transition in the data center a very similar type of spin-in which we had very similar terms to. If you watch our track record here, this is the fourth time this team has built what we hope will be a world class product covering one of the elements (of network programmability)."
Chambers said the product Insieme is developing will complement and not overlap with Cisco's existing products. Cisco identified the opportunity in software programmable and application aware networks a year ago, he said.
Cisco will address the SDN market in a "three- or four-pronged approach": hardware and software, virtualized and physical, intelligent ASICs and intelligent software. He did not delve into specifics or details on which approach Insieme will take, or if it will be based on OpenFlow, the popular open source protocol and API for SDNs.
An Insieme recruiter's letter said the company is building a system to support the OpenStack cloud framework and distributed data storage.
"The end goal, what the people want, is to have programmability, flexibility, starting with massively scalable data centers," Chambers said about Insieme's work. "The way that we will approach it is what we've always done on a market opportunity. It's too early to say which way it's going to go. It's clearly in its early stages (and) too early to say even if it will work architecturally."
Chambers' comments were the first time Cisco's confirmed or even commented publicly on Insieme. Word leaked out about the Cisco spin-in last month.
Since then, the industry's been abuzz about what Insieme is developing. The latest speculation centers on an OpenStack-adherent high-capacity modular switch optimized for 100G Ethernet and storage, aggregating scores of software programmable top-of-rack switches controlled by a Layer 4-7 SDN appliance.
Chambers dismissed a widespread belief that OpenFlow combined with merchant silicon would commoditize network hardware to the point of damaging Cisco's high profit business.
"For 20 years ... every year there was a new competitor, a new technology, a new area that was going to commoditize segments of Cisco switching, etc.," he said. "And every year we tend to focus on the market transitions as opposed to the competition. And we have a healthy paranoia. But anything that loads networks and anything that is specific to programmability on the network and distributed intelligence in it plays right to our sweet spot. So we view this as a possible opportunity and I say possible because, if you ask everyone to define OpenFlow and software-defined networks, everyone would have a different definition. It is truly embryonic. We're just going to be very aggressive in terms of covering our bases just as we've always done before in areas of interest to us. If you asked me a year ago I might have said this is an area where I don't think we have our strategy together on. Today I feel pretty comfortable about where we're going to go and it's more (a matter) of execution."
It's also been noted, right or wrong, that Cisco's spin-in strategy engenders resentment and bitterness among Cisco engineers who are not selected to participate in the start-up's operations, and competition and dysfunction among business units. Chambers intimated that Cisco addresses new market opportunities by any means possible while also working on consistency in product development.
"We have had the courage to go after new product arenas through a combination of internal innovation, start-ups, spin-ins, acquisitions, and there's a reason out of the 18 major product families we focus on we're No. 1 in 11, and No. 2 in four or more," he said. "It's that ability to do the combination. One of the things that (we) insist on is you always make the right decision for the company. In terms of opportunity, there is so much opportunity in switching, data center, in campus, wireless and security - more than we can all address. What is Cisco's biggest challenge? Prioritization. There's more opportunities than we can balance. And if you're operating in a conservative market and expensing appropriately with that and going to give your team a raise, that is the biggest issue (we) have to balance within it.
"What has changed, however, is we used to develop products in silos," Chambers continued. "And actually more of our groups competed internally than externally. We used to develop individual families within Catalyst and Nexus which actually competed. That was OK when you competed against start-ups or players that did it the same way. It's not OK when you play architectures. So we're beginning to bring common ASICs together, common software together."
Read more about data center in Network World's Data Center section.