One of the main goals of SDN (software-defined networking) is to make networks more agile to meet the changing demands of applications. A new Silicon Valley startup, Apstra, says it has an easier way to do the same thing.
Rather than control the guts of individual network devices through software that makes them more programmable, Apstra says it can deal with those devices as they are and shape the network from a higher level.
The result is a new approach that might let IT departments bypass some of the complex technologies and politics of SDN and still make their networks more responsive to users’ needs. It’s due to go on sale by August.
Network programmability has been a mixed blessing, Apstra Founder and CEO Mansour Karam said.
“The network engineer needs to become a software developer in order to take advantage of this programmability layer at the device level,” he said. “That is an unrealistic expectation.”
Apstra’s aim is to let them be network engineers but not use the traditional tools of networking like CLIs (command-line interfaces).
The company approaches the problem of agile networking from the top down where SDN has come at it from the bottom up, Karam said. SDN initiatives like OpenFlow take the functions of network devices like switches and figure out how to realize those in controller software. Apstra looks at what an organization wants its network to accomplish and figures out how to make the switches do that, he said.
The result is simpler and more agile, the company claims. But it can also coexist with classic SDN, current resource management systems like Infoblox, and network telemetry platforms like Cisco’s recently announced Tetration Analytics, Apstra says.
The company’s product, called Apstra Operating System (AOS), takes policies based on the enterprise’s intent and automatically translates them into settings on network devices from multiple vendors. When the IT department wants to add a new component to the data center, AOS is designed to figure out what needed changes would flow from that addition and carry them out.
The distributed OS is vendor agnostic. It will work with devices from Cisco Systems, Hewlett Packard Enterprise, Juniper Networks, Cumulus Networks, the Open Compute Project and others.
AOS takes advantage of APIs (application programming interfaces) for network devices that didn’t even exist until a few years ago when networking began to open up, Karam said. It can also work with Linux-based container environments.
The system uses real-time telemetry that can detect and show whether the network is carrying out policies as intended.
That visibility will be critical for network engineers to embrace a system like AOS, said IDC analyst Brad Casemore. Many businesses want network automation, but networkers are wary of it because they trust the tools they’ve always used, like CLIs. If they can closely monitor what’s going on, they’ll be more likely to trust automation, he said.
That’s part of what’s driving the current trend in network telemetry, which is producing other systems like Cisco’s Tetration Analytics and Voyance from startup Nyansa, Casemore said. They don’t all do the same things, but they aim to satisfy a hunger for information about what’s going on in increasingly complex IT environments that combine elements like cloud, virtualization and mobility.