Success requires a strategy that combines AI-Native and cloud-native approaches
Artificial intelligence (AI) can deliver transformational change in today’s networks with a variety of benefits. But a new Foundry survey also uncovered variations in how IT leaders are implementing AI in their networks, with some leveraging a “bolt-on” approach that may not deliver the desired results.
Why IT leaders are prioritizing AI in networking
The survey shows that IT leaders are all-in on the idea that AI can help them address a long-standing struggle with enterprise networks: making day-to-day management of networks easier. That, in turn, promises to free up IT to spend more time on strategic initiatives while maintaining a superior experience for end users.
Specifically, respondents are using (or hope to use) AI in networking to prioritize data-driven decision making, bolster compliance and risk management, enhance quality of service (QoS) and user experience, and improve network reliability and uptime.
When asked to pick their single most important objective, “improving network reliability and uptime” was number one, chosen by 17% of respondents.
That makes sense to Sharon Mandell, CIO for Juniper Networks, who argues that “you don’t get to do the cool stuff in IT until the core functions work well enough not to be a distraction.”
“Cool stuff” includes focusing on digital transformation efforts, which was the number one project respondents said they’d spend more time on. Close behind: data analytics and business intelligence projects, as well as cybersecurity.
But those time savings will only materialize, Mandell continues, if companies successfully deploy AI in their networks, a job that is not without its challenges, such as
safeguarding the network against AI-specific threats, ensuring the long-term sustainability of AI initiatives, allocating resources effectively between AI and other initiatives, and keeping up with the rapid pace of AI networking technology evolution.
How to get the most from AI in networking
Overcoming those challenges requires a specific approach to AI implementation. However, the survey turned up stark differences in respondents’ approaches.
Twenty percent favor a bolt-on approach, where they add AI solutions to existing networks without significant infrastructure changes. More than a third (35%) say they’re taking an “integrated” approach, which involves redesigning networking infrastructure to fully integrate AI capabilities. Slightly more (37%) favor a hybrid approach that uses a mix of bolt-on and integrated AI solutions.
However, those using a bolt-on approach will inherently struggle to incorporate AI into their IT infrastructure and, therefore, have a hard time taking full advantage of AI’s transformational benefits.
According to Mandell, a more effective approach would build around AI integration as a core component. Juniper Networks’ cloud-based Mist AI platform, for example, uses real-world data to recognize network issues as they’re developing and address them before they result in performance problems or downtime—with no aid from a network administrator.
It’s difficult to bolt on that ability to deliver data to the AI engine and receive instructions from it, Mandell says. “That’s why the cloud-native piece is important. It’s a combination of AI-Native and cloud-native that makes it work.”
