Building the Network of the Future – Powered by AI

The network is the foundation of nearly every piece of every business today. Nearly all employees are online, on company-supported and their own devices. The equipment their businesses rely on is networked too, from basic production tools to advanced manufacturing robots. Even the physical devices that support them, like door locks and lighting switches, are online.

All these devices create data flows distributed among multiple connection types, and between different public and private clouds. To support this increasingly complex, connected infrastructure, the IT industry must change. The manual approach to managing networks cannot keep up. Automation and analytics tools must come together. When they do, businesses will have not just new tools to run their networks, but new foundations on which to build new services and products.

To help make sense of the converging changes in business and technology, we have created the 2020 Global Networking Trends Report. Its predictions and insights come from a new survey of over 2,000 Cisco customers, combined with data from our annual Visual Networking Index, reports from analysts, and interviews with our most experienced over-the-horizon thinkers.

AI Changes Everything

The core technology for this convergence: Artificial intelligence (AI). Only by using machines to augment human skills will we be able to keep up with the complexity and scale of next-generation networks.

The report shows that networking has taken its first steps into a new realm, the ultimate impact of which is difficult to forecast. The disruptive technology that is underpinning this revolution is AI, and it is having an impact on every arena of IT, including security, mobility, user experience, and IT management.

Critically, the available AI power we have is increasing at a rate far outstripping Moore’s law, because AI benefits from the multiplicative growth of several fields: computing power, which Moore’s Law predicts; the improvements in AI science and algorithms; the massive growth in the production and curation of high quality data, which is feeding the machine learning algorithms; and in the inevitable increases in network speed and reliability, which is what connects data to computing power.

OpenAI observes that from 2012 to 2018, the amount of available AI compute power for a training run increased 300,000 times. This capability is available not just to the largest and richest companies. Thanks to cloud services and ubiquitous high-speed networking, it is becoming available to almost every business. This new AI capacity will have a profound impact that we cannot predict.

But we can begin to prepare for it and to leverage it.

Planning for the Unknown

The rapidly learning and evolving infrastructure of tomorrow will not be manageable using yesterday’s tools. Even today, as confirmed by our Report’s survey respondents, networks are becoming unmanageable by traditional means. The changes are taxing IT personnel who are working hard simply to maintain status quo or make small improvements.

At the same time, our networks are increasingly capable of providing rich data that can be used for IT management, security, and for business operations. But getting the value from that data is the challenge: Seeing more data doesn’t by itself lead to the best outcomes. AI makes analytics more immediate and relevant, and helps to provide relevance to business, security, and IT operations.  For example, an AI-enhanced network management tool can prioritize alerts based on what it determines to be normal network behavior, not just based on static thresholds. This capability exists today: Read Cisco AI Network Analytics: Making Networks Smarter and Simpler to Manage.

AI will also help our customers take advantage of software-defined networking (SDN). Today, a majority (64%) of our survey respondents said they have deployed SDN, however only 22% of customers surveyed use AI capabilities for network assurance. But in the next two years, 68% of the customers surveyed project that AI will be feeding them predictive insights and proactive remediation to help them take full advantage of these networks.

Automation and AI is beginning to impact every area of networking, and it is intent-based networking (IBN) that that can bring the domains together. As our Senior VP Ravi Chandrasekaran explains, in The 7 Pillars of Intent-Based Networking, being able to manage an entire network fabric through a unified, software-based controller opens up opportunities to create networks that support end-to-end, user-centric policy management. The biggest initial impact is in security, where intent-based networking can help create (and manage) segmentation on networks to keep different types of devices, users, and applications isolated from each other.

Only 4% of the customers we surveyed consider their networks to be intent-based today. In the next two years, 35% project their networks will be classifiable as such.

The Talent Gap

I recently asked a customer, a network architect for a university, if he was worried about artificial intelligence threatening his job, or the jobs of the team of admins he managed.

“No way,” he told me. “We have so many projects we want to do right now that we can’t, because we spend so much of our time doing basic management.” He said he badly wants to do more than just keep the network going. He, and the people on his team, are eager to use their skills to move things forward, and to run new projects that contribute to their business in significant ways — and not just support the status quo.

Joe Clarke, a veteran of Cisco for 21 years and a distinguished engineer here, believes that by 2025, 75% of networking teams will spend two-thirds of their time creating new value for businesses.

But that doesn’t mean the transition will be easy, or automatic. We need to work on helping network operators make the change to software-controlled networking. Education in network programming and business operations falls to all of us who run IT departments. At Cisco, we are releasing new DevNet programs to do just this.

Since there is so much more for network pros to do, and so much new value yet to be realized from networks, even dramatically increased automation and analytics in networking will be good for IT.

The first generation of networking kicked off a massive amount of new productivity, driving business growth and changing the way the world works for everyone. But that is not the end of the story. AI, automation, and advanced analytics will unleash a new wave of innovation. What exactly that will look like I can’t predict. That’s for network operators in the field to figure out — the ones who are learning how to blend their deep technical expertise with new business acumen.

Be sure to read the new 2020 Global Networking Trends Report.