A survey of IT and networking professionals has found that almost all (97%) of them feel that the use of artificial intelligence (AI) and machine learning (ML) to drive automation in software-defined wide area network (SD-WAN) environments is an important consideration. For some, it is seen as critical. The research reveals that AI technology is expected to drive automation and operational efficiency in complex SD-WAN environments.
The survey of 374 representatives from organisations in the US and Canada involved with networking technology was conducted by TechTarget’s Enterprise Strategy Group (ESG), which notes that SD-WAN environments will need to become more dynamic over time as IT becomes highly distributed and more complex.
The survey shows that network operations teams recognise the need to be more proactive and accelerate mean time to detect (MTTD) and mean time to repair (MTTR). AI, ML and automation are expected to help. Of the networking professionals surveyed, 40% identified anomalous activity detection, 39% cited predictive analytics for early issue detection, and 39% named accelerated troubleshooting as the most important features of these technologies for their SD-WAN environments.
Other uses for AI will be to provide recommendations, optimise performance and, once fully trusted, automate remediation without manual intervention. Given the increased risk of a larger attack surface, the analysts at Enterprise Strategy Group note that it is promising that organisations are planning to leverage the SD-WAN environment in these ways to enable faster detection.
Network equipment providers add AI capabilities
Given that networking professionals appreciate the benefits AI can offer network operations, network equipment providers have been busy adding AI and ML to their product portfolios, extending artificial intelligence for IT operations (AIOps) to support network operations.
For instance, in February, Cisco opened Cisco Live 2024 with the launch of what it calls AI-enriched networking, security and observability, designed to give businesses the visibility and insights they need to connect and protect their entire digital footprint and build digital resiliency. The company claimed it was uniquely positioned to revolutionise the way infrastructure and data connect, protect businesses of all sizes and tackle its customers’ core challenges.
In April, Extreme Networks unveiled AI Expert, which it says has been built to pull data from the network and beyond to improve performance and operational efficiency. AI Expert combines data from applications and devices across the network to establish intelligence on performance and experience. Extreme says the service will curate enterprise data to provide insights, automate operations and create alerts when it detects anomalies such as network overload, degradation or Wi-Fi dead spots, among others.
According to Extreme Networks, AI Expert is designed to turn insights into expertise and actions, recommending preventative actions and network optimisations based on business key performance indicators. Extreme creates suggestions and best practices to troubleshoot, resolve or proactively address issues.
In June, Juniper Networks unveiled a new product offering, designed to use AI for networking to drive even more value to enterprise WAN environments offering assured SD-WAN experiences with proactive AIOps. Marvis Minis, Juniper’s digital experience twin built to improve network operations, has been extended to SD-WAN. Marvis Minis is able to diagnose real authentication issues without requiring users or devices.
According to Juniper, this enhancement means WAN speed tests can be continuously run to verify link speeds and take action proactively if problems are detected, without users having to be present.
Juniper’s WAN assurance product can now proactively capture packets at the time of a bad incident to help identify and fix hard-to-find issues, avoiding expensive and time-consuming site visits. Finally, new application insights offer network operators a user-friendly visualisation of the traffic traversing the SD-WAN.
GenAI helps network admins
One of the most promising areas for deploying AI beyond AIOps in networking is the use cases for generative AI (Gen AI).
It is largely recognised that there is an IT skills crisis. In an article published on SearchNetworking, a sister title of Computer Weekly, John Burke, chief technology officer at Nemertes Research, wrote about how limited IT budgets are leading to a skills gap in networking. He pointed out that people entering the IT profession tend to focus on building skills that can be applied in more general IT roles rather than developing expertise in network operations.
Rather than trying to hire network managers with the right skillset to perform the administrative tasks they will be required to undertake when managing complex corporate IT network confgurations, GenAI could help less experienced IT staff or IT professionals who do not work in IT networking to manage networks effectively.
When generative AI reaches a sufficient level of maturation, it could help network teams automate routine tasks, respond to incidents and account for the reduced workforce, among other benefits
John Burke, Nemertes Research
Burke believes GenAI is evolving into a versatile technology that could eventually support many network operations tasks. “When generative AI reaches a sufficient level of maturation, it could help network teams automate routine tasks, respond to incidents and account for the reduced workforce, among other benefits,” he says. In effect, the GenAI tool acts as a network administrator copilot.
One way in which GenAI could be deployed in networking, according to Burke, is to help burdened network teams create more human-readable and complete documentation of their networks. For instance, networking professionals could use a GenAI tool to read configuration and inventory files, network mapping data and other notes they have already developed. The tool would then generate full written descriptions, even diagrams in some cases.
If network teams layer GenAI – with its aptitude for natural language – onto machine learning AI tools, Burke believes they may be able to handle increased workloads, even as enterprise network staff levels decline.
As an example, he says that once properly trained on the configuration syntax of different network tools, GenAI could help networking staff create network policies. “If network administrators input verbal descriptions of the network intent into a GenAI tool, the tool can generate commands to implement those intents,” he says. “The same is true for the reverse – a GenAI tool can look at configurations and create a description of what the network will do, and network professionals can compare the output with the intent.”
Taking this further, GenAI could also be used to review configurations, which would assist IT professionals when they conduct network audits.
There is also plenty of talk about how GenAI can help programmers. For Burke, GenAI could provide program stubs, write structure, check syntax and offer feedback to help networking professionals create network scripts. However, he adds: “Network engineers shouldn’t immediately use code that GenAI tools provide without question. GenAI can give network teams an advantage on a project, but they should still check, modify and complete codes before execution.”
The future of AI in SD-WANs
The AI-enabled functionality now being made available in network administration tools shows that the industry has recognised the complexity of corporate networks supporting highly distributed enterprise IT environments. These advanced tools offer the potential to make such networks more manageable.
The challenge the industry faces is that networking complexity is set to increase, which will put even greater demands on networking professionals. While industry experts do not anticipate the emergence of fully automated network management, any help an AI assistant can offer is likely to be most welcome.
