Why SDN all-stars are heading to Brocade
Over the past two years, the company has lured a handful of industry All-Stars to work on software enabling its networking portfolio, including Fibre Channel storage-area network switches, and Ethernet switches and routers. The most recent hire is Michael Bushong, who jumped from start-up Plexxi to Brocade late last year to run product management.
It started two years ago when Brocade snagged Distinguished Engineer David Meyer from Cisco. In between Meyer and Bushong, Brocade also attracted high-level and highly visible software engineers -- Benson Schliesser, Tom Nadeau and Colin Dixon -- from IBM and Juniper Networks to help shape its software and SDN strategy.
Some say Brocade acts as a nimble software start-up funded by its bread-and-butter Fibre Channel SAN business, which owns 70% of a $1.8 billion market. The acquisition two years ago of Vyatta, an open source networking stalwart, didn't hurt either.
+ MORE ON NETWORK WORLD:Brocade seeks SDN edge with early OpenFlow 1.3 support +
"Brocade was a place where I could spread out a bit," says Meyer, who at Cisco focused on examining the role of OpenFlow and SDN within the enterprise and service provider markets. "One of the things that happened to me at Cisco was that it got too small for me in terms of its vision and that was a concern for me."
Meyer's vision, which he shares with Brocade CTO Ken Cheng, is that the future of networking is all about software. As Meyer traveled around to evangelize this vision and the opportunity for it at Brocade, it became a siren song to Schliesser, Nadeau, Dixon and Bushong.
"They have gone after known individuals in the area, and working at Brocade is attractive compared to doing SDN at, say, Cisco, where whatever it is it isn't strategy No.1," says Peter Christy, director of networking research at 451 Research.
It also didn't hurt that at Juniper, where Schliesser and Nadeau were, SDN was creating a rift internally; or that IBM, home to Dixon, was selling offpieces of its SDN business and assets. Another attraction was that Brocade was heavily invested and involved in OpenDaylight, the vendor-initiated open source SDN project.
Brocade became one of, if not the first vendor to commercially offer an OpenDaylight SDN controller.
"It helps that Brocade has leveraged open-source projects, including OpenStack and OpenDaylight, so extensively," says IDC analyst Brad Casemore. "Many [Brocade migrants] were involved with the open-source projects, and they must have felt that Brocade's commitment to open source was genuine and substantive, and that the company will be likely to provide the resources and support they'll need to continue their work."
The Vyatta difference
One of those resources is Vyatta. Brocade acquired the open source networking nemesis of Cisco to pursue new market opportunities in data center virtualization, public cloud, enterprise virtual private cloud, and managed services, and bolster its overall software networking capability.
Vyatta is a key piece of Brocade's Network Functions Virtualization (NFV) offerings for cloud and service providers.
"The company's approach to letting Vyatta develop autonomously, effectively inoculated and distinct from the company's established hardware business, was a smart move," says IDC's Casemore. "As networking value migrates to software, Vyatta is well placed to benefit by providing a platform for virtualized network functions."
Meyer says Vyatta, OpenDaylight, OpenStack and OpenFlow were merely appetizers to the main course at Brocade: a significant growth opportunity in software based on strategic importance, autonomy, technology agnosticism, and work on machine learning -- machines that use data to make predictions or decisions rather than requiring explicit programming for instruction.
"When network people get to work on machine learning, if that's what they want to do in the industry, then that's a great thing," Meyer says, referring to his own experience at Brocade, a $2.2 billion company. "There's great opportunity here not only in networking but in what a technology company might do."
Machine learning likely has more upside than Fibre Channel SANs, the technology on which Brocade built its business. Though Fibre Channel SAN switches are still close to a $2 billion business, it's been lumpy: the market grew 6% from 2011 to 2012, but declined 6% in 2013, according to Dell'Oro Group, which expects 2% growth for 2014.
Brocade has gained about 2 percentage points of revenue share in each successive year between 2011 and 2013. In the third quarter of 2014, Fibre Channel switch revenue grew 1% sequentially and 2% annually for the company, while Cisco's grew 51% sequentially to $171 million based on pent-up demand for 16Gbps products, according to Dell'Oro.
So the aggressive moves into machine learning specifically and software generally are for long-term survival, notes 451's Christy.
"Brocade has a strategic problem because the market life of FC networking is limited," he says. "The CEO [Lloyd Carney] has concluded that SDN is Brocade's best chance of doing something of strategic impact."
Meyer says the software efforts are not at the exclusion of the traditional business, but integral to it and vice versa.
"We envision SDN as sort of a network thing but I call it CSNSE: Compute, Storage, Network, Security and Energy," Meyer says. "You have to have all of these things working together in order to optimize them the way our customers want to."
Sounds like a software-based hyper-convergence project in the making at Brocade. Is that where all of this is leading
"All of the stuff we're working on -- Vyatta, [the OpenDaylight SDN] controller, pieces of OpenStack -- are all pieces of that story," Meyer says. "Building out our expertise in the SDN and OpenStack space is kind of crucial to all of this."
So is Linux-based container technology, which is also on Brocade's radar but perhaps on the periphery. Meyers says the Vyatta folks have their heads down developing use cases for their software as they evaluate a potential fit for container development tools, which allow apps to be ported across Linux hosts without requiring a full-blown virtual machine.