Cisco says its 35-year history will give it an edge in AI

Cisco has introduced new AI software for network analysts that helps differentiate a general issue from a potentially damaging threat to a network.

The new tools are designed to help human IT teams wade through the large volumes of alerts generated by analytics in Cisco's DNA Centre and identify the ones that can really hurt a business.

The new software enhancements aim to improve the accuracy and reliability of threat detection while being able to identify trends and root causes too.

Cisco's DNA Centre users surface around 8,000 'anomalies' every day, according to the firm, with many of these being superficial errors that do not require immediate attention. In a test case, the new system was able to whittle down the 8,000 anomalies to around 300 personalised core issues that need to be remedied.

"That's a 75% reduction in alerts, it's a huge, huge increase in IT efficiency, so now I can pay attention to those 303 that are really unique to me and unique to my environment," said Scott Harrell, SVP and GM of Cisco's Enterprise Networking Business. "It makes me much more efficient, makes me much more able to provide a better service to my users."

When a threat detection product learns what's normal and what isn't, it can deliver legitimate, actionable alerts to IT teams, sidestepping the mammoth 8,000 alerts given in the past, according to Cisco.

Machine learning could also provide organisations with the ability to monitor trends on a network autonomously which means threats can be anticipated and fixed before they cause damage to a network.

"One of the cool parts about machine learning is that I can take seemingly uncorrelated pieces of data and I can start to actually provide insights as I watch these pieces of data move in concert," said Harrell.

"I can start to become predictive about when I see certain problems arise [I can see] the precursors that led to those problems and start to flag those well in advance before they ever became a problem for the network administrator," he added.

Cisco is somewhat late to the party when it comes to analytics tools, however, its CEO believes the new tools give it a unique place in the market.

"We're the only company that has technology that you can integrate across the datacentre, the campus, switching to wireless, enterprise router platform," said Chuck Robbins, CEO and Chairman at Cisco. "[We also have] a robust security portfolio, including cloud security gateways and you can see many of our competitors doing acquisitions in this space to try and extend their footprint because they know that customers want this.

"I think our advantage is we're going to have more seamless architectures and we have technology in all of them so that should give us quite the head start."

Cisco also thinks that its 35-year history in the market has allowed it to collect one of the richest networking datasets, something which is fundamental to creating an effective AI product.

"It all starts with the data and Cisco has one of the best datasets than anybody in the industry," said Harrell. "We've been using machine learning, we've been using AI for decades now for things like security - we block over 20 billion threats a day in our security portfolio."

"So, we have one of the richest datasets in the world and we're going to apply that to networking," he added. "We have a huge high volume of data from every dimension you can imagine and that's really key - it's not just the volume of data, but the diversity of data and the quality of that data."