Automation, AI and ML for Network Management: It’s Here, Are Organizations Ready?


Automation, AI and ML for Network Management: It’s Here, Are Organizations Ready?

From physical to virtual to cloud infrastructures, networks are getting more diverse. For network teams, managing across hybrid environments is a growing challenge. On top of that, increasing adoption of SDN and SD-WAN technologies are adding new virtual, overlay, and underlay constructs and elements, all making the network stack even more difficult to holistically understand. These factors make accurate network visibility and effective network management harder than ever to achieve.

What’s the answer to managing today’s networks? Automation, based on advanced data integration and analytics, according to many in our industry. Humans and manual processes can no longer keep pace with network innovation, evolution, complexity, and change. That’s why we’re hearing more about “self-driving networks,” “self-healing networks,” “intent-based networking,” and other concepts which aim to apply artificial intelligence (AI), machine learning (ML), and automation to support modern network operations.

Collectively, this shift is being referred to as “AIOps” (i.e. artificial intelligence for operations). That is, an approach that leverages multiple sources of real-time and historical monitoring data, adds contextual enrichments, applies AI/ML to recognize patterns and anomalies worthy of actions, and automates corrections, however and wherever practical. Such systems can operate around the clock, at any hour of the day, and respond long before humans can manually sift through vast swaths of data to make discoveries themselves.

The potential benefits of AIOps, which include significantly improving responsiveness and effectiveness, are what makes the approach highly appealing. For instance, by better leveraging existing, available network data, AIOps can reduce the most time-consuming manual troubleshooting and analysis tasks, so that networking teams can focus more on growth rather than firefighting. AIOps concepts can also be applied to enhancing awareness and accuracy for better network engineering, capacity management, and cost control. Further, AIOps includes integrated automations to address and correct issues, ultimately ensuring fast, secure, and performant delivery of digital services.

With AIOps already taking shape within other technology groups, Kentik conducted a survey at the recent Cisco Live U.S. 2019 conference to gauge progress. This report, “The State of Automation, Artificial Intelligence, and Machine Learning in Network Management,” aims to help our fellow networking peers better understand and prepare for the adoption of automation, AI, and ML within their own organizations. A few of the report’s key findings include:

  • Network automation is taking shape. 85% of respondents said their organization has one or more types of automation, and yet only 27% of respondents said their organization is “extremely prepared” or “very prepared” for full automation.

  • The energy sector leads the network automation trend. Healthcare and government are behind the curve. 

  • Networking processes like compliance and incident response are least likely to be automated. 53% of respondents are using automation for network configuration - the only area to receive a majority response. 

  • Machine learning is growing in importance for network management, regardless of who you ask. Up 20% since our 2018 survey, 65% of respondents said that machine learning is now extremely important” or “very important” for network management.

  • And much, much more.

Be sure to download the report and let us know what you think at the next NYNOG event. If you’d also like to know more about Kentik, check out two key resources from us, one made for service providers and one made for enterprises


Michelle Kincaid, Director of Communications