Cisco took the “first pivotal step” in delivering its Predictive Networks vision to introduce ThousandEyes WAN Insights that aim to proactively forecast network conditions and provide recommendations to optimize SD-WAN performance.

The new ThousandEyes product “is the first example of true predictive network intelligence that doesn't just leverage machine learning (ML) to understand what's happening and what might be wrong now, but actually predicts outages that have yet to happen, and will allow you to remediate them in real time,” Todd Nightingale, EVP and GM of Cisco's Enterprise Networking and Cloud business, said during the Cisco Live event.

It will be initially available for Cisco SD-WAN customers. “SD-WAN just was the most natural starting point because it's being massively deployed in every organization right now and it is making the internet supercritical,” Mohit Lad, VP and GM of Cisco ThousandEyes, told SDxCentral, adding that due to some organizations not deeply understanding this fairly new technology, this is where they can make actionable recommendations.

Cisco acquired ThousandEyes and its internet and cloud intelligence platform in early 2020 in a deal valued at $1 billion. Earlier last year, the company extended ThousandEyes network performance monitoring agents to its Viptela SD-WAN appliances.

ThousandEyes offers a view of the internet and WAN insights connect this view to the WAN, Lad explained.

The ThousandEyes WAN Insights offers policy recommendations and path optimization guidance, and alerts IT team issues before they happen, Cisco claims.

The prediction capabilities came from the networking giant’s newly-announced Predictive Networks engine, which gathers data from the vendor’s vast telemetry sources, uses ML to learn the pattern and build a model, and then predicts application issues based on the model and provides solving options to improve network user experience.

Learning from past behaviors, it can forecast disruptions and "it fits really nicely with the mission that we all have of shifting from reactive to proactive,” Lad said, referring to the new product. “It's using data to design better networks, using technologies, such as SD-WAN with the ThousandEyes datasets to make the right decisions” and help better collaboration among internal and external teams.

Currently, the product only provides long-term recommendations focusing on more systemic issues. “It can show you historically how the recommendation would have played out for the last seven days, and it shows you what a test recommendation would have taken you to,” he added. “So you build that trust because you look at the history and you can decide whether you want to do it or not. And then it's a clear recommendation. We're not automating it right now.”

For now, there will be a human in the loop and recommendations have to be longer term until it gains the trust from customers to allow it to automate the remediation, Nightingale said.

“As we start to close the loop and start to do direct automatic remediation, then we can start to put [automation] on the horizon,” he added.

As IT teams accept those recommendations more frequently, eventually they will be open to pushing the remediation automatically, Nightingale expects. “I hope that we will gain that level of trust and have people closing the loop and do fully predictive remediation in a fully automated way … I think it's going to happen sooner rather than later.”

But the platform is launching with predictive insight first because ThousandEyes wanted to roll out the product early and also learn from customers in order to figure out the automation options, Lad said.

“People automate things with ThousandEyes even today, because we have APIs and all that, so they can build their own automation or use Ansible [and similar] frameworks to build automation around changing routes and recovery,” he added.