The Intent of Intent-Based Networking


By: R. Scott Raynovich

Following buzzwords is no easy task. Not only do you have to catalog, understand, and define new buzzwords -- but then you have to negotiate the tricky vendor landscape of how they are used and abused and who got there first.

Intent-based networking is the hottest new buzzword in the networking space -- so hot that Cisco recently took the entire networking marketing strategy and portfolio it's been working on for five years, then threw it all away (I'm referring to the fact that they did not even mention Application Centric Infrastructure) to rebadge it under the banner of intent-based networking and Distributed Networking Architecture (DNA). Cisco is also calling this "intuitive networking."

Is any of this new? Or is it just marketing hocus-pocus. That's the buzz in the industry.

Some of Cisco's smaller competitors, such as Apstra, a Silicon Valley startup that has a legitimate claim to have started the concept of intent-based networking, have been irked by Cisco's marketing engine. First, there's the issue of whether Cisco is actually doing what intent-based networking is. Second, there's the idea that Cisco is trying to claim "industry firsts" and hijack the intent-based theme, after somebody else got there first.

Intent Washing

I recently caught up with Apstra CEO Mansour Karam. He said that intent-based networking is so hot that now everybody is trying to take advantage of it. He calls this "intent washing."

"We were told this would happen," said Karam. "Intent washing is when every vendor rebrands what they do with SDN."

That's a great point. Isn't intent-based networking what software-defined networking was supposed to be in the beginning? The idea is that you have a network of open, compatible networking devices that you can program with "intent" -- tell it how to adjust to changes in requirements -- and then let the network run itself with automated software.

Apstra, for example, defines intent-based networking as setting up software control that can anticipate and resolve networking problems in real-time -- with automated responses. For example, in the old-school world of networking, individual network engineers have to monitor network resources and manually configure the network to adapt to changing conditions. Intent-based networking automates this, by dynamically adjusting the network in real-time -- without human involvement.

But this requires a lot of sophisticated technology. The software must understand and interoperate with many types of hardware, pull data and telemetry feeds from many sources, make sense of that data, and then intelligently send commands to configure the hardware.

Whether this can actually be executed -- and work -- comes back to the strength of the data and analytics in the software, as well as the ability to connect to as many devices as possible. This requires lots of devices with open APIs.

"You cannot be an intent-based system if you don't have telemetry and you aren't monitoring this telemetry in a closed loop," says Karam.

D-Day in Networking

One interesting thing about intent and the battle for marketing prominence in the field is that the concept did not originate in Silicon Valley or even in the technology industry -- the concept of intent-based command originated in military and has been a staple of business leadership jargon for a while.

Intent-based command comes from the evolution of a central command system to a more distributed one. After decades of evolution, intent-based command is a common term in military manuals and doctrines -- just consult the US Army's summary of Commander's Intent and Concept of Operations, for some military history.

In 2010, the Harvard Business Review published an interesting article called "Manage Uncertainty with Commander’s Intent," which described how Commander's Intent was used during D-Day to adapt to changing conditions in the battle after many original plans failed on Day 1. As many HBR articles do, this probably helped get the ball rolling with intent being in technology marketing circles.

Intent is also common in leadership jargon -- as in, "intent-based leadership." A guy named David Marquet has apparently done pretty well with that theme, rising to the top of the Google results. The idea is to not tell people exactly what to do -- just give them a goal and let them figure out how to get there.

So -- be prepared for the era of intent. Will we have intent-based marketing? Intent-based driving? Intent-based diets? I'm sure it's all coming soon.

Apstra Defines Intent

So what should we do about it in the networking market? Is Futuriom going to start issuing reports on "intent-based networking"? I'm not quite sure yet. The way I see it, intent-based systems are SDN 2.0.

As Cisco steps up to make all sorts of claims in its new marketing push, it will be important to define the functionality of intent-based networking and whether it's being achieved. So maybe it's important to have a definition first.

Apstra's Karam helped me further define what it is. The key components are:

1) Telemetry and data gathering
2) Software-based analytics
3) APIs and hardware interoperability
4) Networking automation.

As we look for the move to an intent-based model in networking, you could evaluate how well it's going by whether we're delivering the four elements outlined above. The vendors should be graded on their execution on points 1-4, and how well they are delivering.

Point #3 is very critical to the equation. For all of this to work, the networking hardware vendors have to share APIs and move to a more open environment -- and the customers have been very clear that's what they want. This will be one of the more important things to watch to see if Cisco really is intent (ha) on building more open systems, or whether it's just figuring out a clever way to build a new moat.