If Henry Ford Had Only Made Incremental Changes...
MOUNTAIN VIEW, Calif. -- Open Networking User Group Spring 2016 -- Network operators seeking innovation are caught in a Catch 22: They need to focus on incremental change to protect existing technology investments. But incremental change doesn't lead to business transformation.
"If Henry Ford had worked incrementally, would he have come up with an automated horse feeder rather than the Model T?" said David Cheriton, a serial networking entrepreneur, venture capitalist and Stanford University computer science professor, at a panel discussion here.
But James Younan, UBS director, infrastructure research and development, had another perspective. "Henry Ford didn't have 15,000 existing things to compete with. He had a clean slate," he said.
Younan added: "The technology part is considerably more easy than the business process changes required." Younan said he has worked at organizations with innovative technology that were completely unable to change business processes.
And sometimes the business processes are hard to change for a reason. "As we automate, we need to keep some controls in place to make sure something doesn't go rogue and destroy the network," Neal Secher, managing director and head of network architecture at BNY Mellon, said.
Earning trust and buy-in from business managers is essential to change, Younan said. Network managers need to avoid forcing change on business users. "Those things are rarely adopted in a good way," he said.
Greenfield innovation is easier to accomplish than brownfield, Aryo Kresnadi, technical fellow at FedEx, said. Pablo Espinoza, director of network Engineering, Intuit, agreed. "The greenfield is bright and shiny," he said.
Business managers need to be convinced to move to the greenfield, Secher said. "It's about incentivizing moving to the greenfield and making the greenfield a great place to be," he said. Businesses underestimate the greenfield opportunities available. "There are new workloads coming out daily," he said. Business requires hybrid, private and public cloud, and new ways of interacting with the network -- all of which provide new opportunities for greenfield innovation.
Solving business problems is the best way to win support for innovation from business managers. Speeding up service deployment is one good way to do that -- deploying services in an hour rather than 30 days, Secher said.
Network operators need to partner with businesses and show them the value, Secher said. "That's a snowball that will roll downhill, where you have more opportunities to automate more and more."
But business users can be hard to please. "To the user, it's a plug in the wall and it just works," Joe Ferrell, leader and CTO, computing and datacenter infrastructure, GE, said. "They don't understand the detail or want to know the detail. They just want it to perform and be up all the time. Just as when we're at home we want the Internet to be up all the time."
The right people can help drive innovation. That requires fresh eyes looking at problems, without the baggage of CLIs and configuration scripts and 15 years of history, Secher said.
Software development and infrastructure experts need to work together, according to Espinosa. "You won't find those skills in one person," he said.
Younan agreed. Organizations need to break down silos separating storage, compute and network capabilities. "We treat it like it's not a single system," he said. For example, "to deny the fact that to do IP storage I need a network and it needs to be designed properly is just plain ignorance," he said. Compute, network, storage and software all need to be connected holistically.
And they also need to be vendor independent, he said. Networking and storage has lagged behind compute and software in that regard.
But there may be good reason why networking has evolved more slowly, Cheriton said. "I'm a bit of a network bigot," he said. "If the server fails, the server fails, but if the network fails the business is on the line." In other words, maybe the network has lagged because it's so important.
Modern web companies such as Google (Nasdaq: GOOG) and Facebook are architected for change, with gradations between brownfield and greenfield and automated test processes, Cheriton said. Cheriton co-founded Arista Networks Inc. and was an early investor in Google and VMware Inc. (NYSE: VMW).
But that innovation is possible for modern web companies because they don't have the legacy, Younan said. Most large enterprises are dealing with 7,000 to 8,000 applications written many years ago, while large web companies have about 40 apps.
"We have 8,000 Pablo Picassos that are each a masterpiece on their own," Younan said.
— Mitch Wagner, , West Coast Bureau Chief, Light Reading.