How to Make SD-WAN Predictive Automation a Reality

In a study conducted by IDC, respondents ranked “Keeping up with technical and market developments” as one of the top three main challenges when managing an SD-WAN network. SD-WAN adoption across organizations has expanded rapidly, and so have the technology tools to help maintain networks.

As global enterprises continue to expand SD-WAN deployments, which are increasing in scale and size, the need for more simplified and efficient management grows. Also, the underlying connectivity options of SD-WANs can be highly variable, with users, locations, transport options, and applications being fluid. Therefore, management platforms must have all the tools to be adaptable to each specific organization. There are a variety of approaches organizations can take to simplify networks, including intelligent and predictive analytics and automated management actions. Simplifying network operations can be achieved through streamlined management, automated actions, and end-to-end intelligence.

Top requirements to enable SD-WAN predictive automation

Enterprise network managers must consider three key requirements to ensure automation drives value: End-to-end intelligence, analytics-driven prediction, and predictive automation, all rooted on a solid and consistent telemetry foundation.

End-to-end Intelligence

Detailed historical visibility into a specific SD-WAN's traffic and applications is key for driving more insights and analytics that can allow IT teams to make quick decisions that drive network simplicity. To do so, monitoring and management solutions must deliver a top-down and end-to-end view driven by data, enabling in-depth analysis to direct specific management actions. A variety of metrics and reporting tools are helpful for visibility: detailed metrics on WAN link performance both at a macro view and a per-circuit level of latency, loss, and jitter; visibility into the health and performance of individual users and applications; and an intuitive dashboard for displaying the information for rapid analysis.

With this knowledge, IT teams have an end-to-end view of all network processes, enabling data-driven management decisions to achieve simplicity.

Analytics-driven Prediction

To further drive network simplicity, IT teams can deploy analytic and predictive solutions to power infrastructure optimization. To achieve simplicity, a successful analytics engine should capture various aspects of data to produce actionable insights and trigger automated systems if an issue arises. It must focus on both identifying problems and solving them efficiently, lightening the IT team’s workload. This is especially helpful when combined with automation that is integrated directly into the predictive platform since it can recommend changes to achieve optimal performance before any user-impacting events.

An SD-WAN deployment with analytic-driven predictive solutions that automatically adapts to any issues or threats that arise not only drives simplicity but also frees up resources so IT teams can focus on more complex issues.

Predictive Automation

The goal of predictive automation is to establish a perpetual optimization cycle where visibility and analytics detect areas of improvement, make recommendations, and drive changes. Traditionally, network managers have manually evaluated technology options to help predict future network behavior and make changes to ensure high levels of service and experience. With predictive automation, the network can drive this process automatically. By implementing the power of artificial intelligence and machine learning, issues within the infrastructure can be identified, resolved, and mitigated in the future.

For example, predictive automation can be applied to SD-WAN application policy prioritization. By ingesting telemetry, analyzing data, and applying predictive modeling, the SD-WAN system can recognize the probability of meeting the network performance needs of applications across various network paths. It can then either recommend a change in the new policy or be programmed to automatically implement it. Modeling can also be applied to circuit link usage based on historical traffic trends to forecast future circuit usage and recommend changes if appropriate.

Before deploying solutions enhanced by artificial intelligence (AI) and ML capabilities, it’s important for organizations to consider that IT teams may be at varying stages of comfort in embracing advanced automation capabilities. For instance, while an advanced management system may be capable of fully closed-loop problem remediation, customers may want to maintain control in making changes to their networks and leverage guided remediation capabilities instead.

A look ahead for SD-WAN management and automation

By arming SD-WAN networks with end-to-end intelligence, analytics-driven predictions, and predictive automation solutions, IT teams can simplify infrastructure management and assure higher levels of quality experiences for users. IT organizations considering automation-driven SD-WAN platforms must examine options for visibility, intelligence, prediction, and automation to gain a virtuous cycle of perpetual optimization for their network.

JL Valente is Vice President, Product, Enterprise Routing and SD-WAN at Cisco.

Related articles: