Cisco today introduced its Predictive Networks, which gather data from the vendor’s vast telemetry sources, use machine learning (ML) to learn the pattern and build a model, and then predict application issues based on the model and provide solving options to improve network user experience. 

“One of the mechanisms that we believe is used by the brain to learn is a prediction,” JP Vasseur, Cisco Fellow, head of engineering for ML and artificial intelligence, told SDxCentral. “We're far from mimicking the brain, but we're using the ability to learn all the time so that we can make better decisions.”

Cisco's solution is powered by predictive software engines the vendor has been building and testing with 100 networks for the past two years. The company plans to deliver these predictive technologies across all domains as integrated, plug-and-play software-as-a-service (SaaS) offerings, without the need for extra hardware, according to Vasseur.

The predictive capabilities can help IT teams to manage cybersecurity threats, hybrid work, and hybrid cloud environments, Cisco claims. The vendor also aims to address the IT talent shortage with Predictive Networks.

“It's an enormous game because today when [problems] happen, it is too late,” Vasseur said. Then it takes a lot of human resources and time from the IT teams to start troubleshooting and finding a root cause, he added. 

Early customer trials showed the engine is capable of anticipating issues and offering options to avoid them, which will save time and improve the quality of the network, he added. 

Many networking vendors already offer self-healing capabilities in their SD-WAN services. However, Vasseur argues the core of “self-healing” is prediction.

“For me, self-healing is about the capability of the network to fix [the issues] by itself without being told, without the user in the loop,'' he said. “The one piece that makes it smaller [than a closed loop] is the ability to predict.” 

And “the Predictive Networks is one step towards the true self-healing network, where the network will have the intelligence to learn, to predict, to plan and to take some action automatically,” he added.

Vasseur touted Cisco is “in a very unique position” to deliver this predictive engine, since the vendor has a decade of experience designing embedded AI systems and a massive amount of data with high diversity. “This is the first engine that is really leveraging this data and this approach of learning,” he said.

Vasseur called the Predictive Networks “a game-changer,” and said today’s announcement is “the beginning of a journey.” Cisco plans to introduce more services around the solution over the next few months.