Data center networking architecture draws on machine learning, SDN, AI
A comprehensive collection of articles, videos and more, hand-picked by our editors
LAS VEGAS -- As a concept, intent-based networking is what enterprises need to manage a network of the future --...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
one that’s likely to span the data center, public clouds and internet of things. The complexity created by having applications running on these diverse computing environments can’t be tamed with the arcane command-line scripts engineers have used for years to manage networks.
At the Future:Net conference in Las Vegas this week, experts discussed what's needed in a complete intent-based networking (IBN) system, which no vendor has today.
"What people are looking for is autonomous networks," Brad Casemore, an analyst at IDC, said. "We're not there, but that's the direction [the industry] is going."
What engineers envision in IBN is the ability to describe network changes in almost plain English and have those instructions compiled into a language that open, programmable devices can understand. The devices could be virtual or physical and span all computing environments.
What's missing in IBN
Ideally, the IBN system would also have the artificial intelligence (AI) or machine learning to automate changes resulting from routine network events, such as moving or spinning up virtual machines or containers. AI in networking today is non-existent, according to Ratul Mahajan, CEO and co-founder of startup Intentionet, which is building products on top of an open source network configuration analysis tool, called Batfish.
AI applications today do not produce intelligence that's universally useful to network operators, Mahajan said in a panel discussion following his Future:Net keynote. Therefore, defining AI in networking is difficult.
"If it appears intelligent, it's AI," Mahajan said. "If it doesn't, then it isn't."
The closest most vendors come to IBN today are systems that create and distribute policies that tell switches, firewalls and load balancers in the data center how to handle specific types of network traffic. Just building and distributing policies isn’t enough, however.
What's lacking in these products are the analytics that provide network engineers with complete information on the state of every device and also help them formulate a policy that accomplishes what they want. Too often, engineers find out after deployment that policies fall short, typically through network monitoring applications.
"You're not going to have in intent-based networking [today] everything that you want," said Peyton Maynard-Koran, an engineering director at video game publisher Electronic Arts.
Startups are working on a variety of IBN tools. Examples of those companies include Apstra, Inc., Forward Networks and Veriflow Systems. The three software developers gave presentations on IBN at Future:Net.
Network automation footprint grows
Move to intent-based networking to automate manual tasks
Intent-based networking transfers service delivery focus to overall network