Startup Veriflow Networks has landed $8.2 million in series A funding. The A round was led by Menlo Ventures, along with its existing investor New Enterprise Associates.
I wrote about Veriflow Networks earlier this year as it was coming out of stealth. The company aims to prevent network outages caused by operators or breaches by using mathematical validation to ensure that configuration or ACL changes on network devices won’t violate policy, open holes, or cause availability problems.
Essentially, Veriflow grabs configuration information from CAM and ACL tables on network devices, including switches, routers, and firewalls, and builds a model of the network as it’s actually configured. This model lets Veriflow chart how packets and flows move through the network, and predict how changes will affect that movement.
The Value Proposition
I was intrigued by Veriflow’s approach because it can help operators grasp complex networks as they are actually configured, as opposed to an ideal or assumed configuration. That grasp is essential if we’re going to take advantage of things like SDN and network automation.
SDN and network automation aim to simplify operations, but these technologies don’t necessarily reduce network complexity, they just mask it behind layers of abstraction.
And networks are poised to get even more complicated. We’re at the cusp of trends such as microsegmention, containerization (lots of little service instances that need connectivity popping in and out of existence), and the programmable network (where interfaces are exposed to developers and applications on switch and router configurations, network OSs, protocol stacks, and even network ASICs).
Having a tool that can track network state in something close to real time, and be able to run sanity checks against proposed changes and configurations, is going to be essential in a world of decentralized and loosely orchestrated operations.
This isn’t to say that Veriflow is definitely the company to do it, or that its approach is the right one. There are still a lot of challenges here. To be effective, Veriflow would have to be integrated into various tool chains and workflows across multiple silos. And a feedback mechanism has to be in place so that if the system does detect a problem with a new configuration, the right teams can respond to it and resolve it in a timely manner.
That said, I’m pleased to see that smart people are thinking about these problems, and that money people are backing them up with investment.