AI: Causing and solving networking problems
Rentia Booysen, Westcon Director at Westcon-Comstor Sub-Saharan Africa
Artificial Intelligence (AI) is no longer the elephant in the room, and with the much-spoken-about release of ChatGPT followed by Microsoft’s announcement that it’s investing $10 billion into AI, it’s firmly off the “in the future” table.
For networking professionals, there are two factors to consider with the rise of AI. Firstly how will the flow of its traffic affect networks, and secondly, how can they use it better manage their network?
Are you still in control?
Over the past two years, the rapid shift to the cloud has left many enterprise network teams in disarray. In some cases, teams have lost control of their networks as the core of the business has moved from their on-premises environment into a hybrid cloud environment. The challenge presented to the network team is that their traffic still behaves as it should flow to the data centre. Which now requires reimagined network management and workflow automation.
While AI can undoubtedly help monitor networks, it also adds its own stress to the network. Cloud-based AI tools require the network to manage and accommodate heavy traffic between on- and off-premises environments as they shift and move massive volumes of data traffic between these. The reality is that AI is everywhere, in analytics tools, IoT and intelligent edge devices, spam filters, and even content creation tools. As these demand their share of the network, they also create traffic surges and latency issues.
AI for mission-critical networking
Conversely, AI-powered traffic management, network management and monitoring tools are maturing. However, while these AI-infused tools are providing a lifeline to resource-constrained network teams, there is still some scepticism concerning how much control we can really hand over to these systems to help manage what has become an increasingly fragile network. The fear? Potential network disruptions and even further loss of control.
The answer lies in using “explainable AI,” or AI solutions that network managers can still engage with, and the inner workings of which they understand. When a network team understands how the AI has come to its decisions and can use regular feedback from teams on the success of the AI’s findings in boosting or managing performance, trust starts being built.
Power of embracing AI in the network
But beyond the scepticism, enterprise networking has been one of the sectors most aggressively adopting AI and automation. It’s used by networking teams in various network functions that extend to performance monitoring, alarm suppression, root-cause analysis, and anomaly detection. For example, Juniper Mist AI automates network configuration and handles optimisation.
The primary catalyst is that AI can help to improve customer experiences. In a recent article, Juniper Networks Chief AI Officer Bob Friday said, “AI’s ability to adapt and learn the client-to-cloud connection as it changes will make AI ideal for the most dynamic network use cases.”
One example of where AI can help improve the customer experience is the wireless user experience. It can offer insights and better management of a spiderweb of wireless connections resulting from mobile devices and work-from-home use cases. In this scenario, AI provides insight into an environment many networking professionals have lost control of.
Handing over some of the reins to AI
One of the more common applications for AI within the networking realm is its role in search and chatbots. With chatbots and virtual assistants architected using Natural Language Processing (NLP) and Natural Language Understanding (NLU), networking professionals can dig their way out from under a pile of support tickets.
When these bots understand questions posed by users, they can respond with information and recommendations based on the knowledge they have gained from observing the network and insights they have been trained to deliver. It’s a form of client-to-cloud insight and automation where chatbots offer context and meaning to a user’s questions beyond just a yes or a no. And the longer they are in operation, the more intuitive they become.
When using Juniper Mist AI and its Marvis chatbot, a global retail giant has been able to glean insights about what is potentially wrong with its networks and how to fix it. Because Mist AI continuously measures baseline performance, it will automatically sound the alarm if there is a deviation.
Preparing for AI
In a skills-strapped industry, IT and networking professionals must embrace the notion that AI will relieve them of mundane, repetitive chores. They should also know that it would be remiss of any company to expect a networking professional to become an AI expert overnight. What they should be gearing themselves for is the inevitable exposure to AI-enabled devices and systems.
To better manage their networks, networking professionals should identify how they can start to manage these networks with their brains, work in tandem with data scientists, developers, and IT to identify the AI-infused tools they need and begin to use AI in the network more effectively.