What if the network continually delivered the user experience that the network operator intended? What if the network was fully instrumented to generate telemetry data, and based on real-time analysis of this data, the network was automatically tuned or reconfigured so it always operated the way we wanted? That day is coming soon.
Big data, machine learning, and cloud-based service delivery: We've entered a new era of innovation that has the potential to unlock untold business value from the network.
A System that Manages Itself
A network that can automatically deliver—and maintain—the service the operator wants is predicated on having a closed-loop system. Closed-loop systems are commonly used in electronic control systems across many industries. They're designed to automatically achieve and maintain a desired "system state" condition by comparing the actual state with the desired state. It's a fully automatic system in which the action is dependent on the output.
Think of a smart dryer that continually monitors the temperature or wetness of clothes, tunes the heat accordingly, adjusts for unexpected events like the door opening, and shuts off when the clothes are dry. A swimming pool that maintains the perfect amount of water at the right temperature for a dip on a hot summer day. The cooling systems in a nuclear power plant.
Closed-loop systems are commonly used in communications systems, too. IP routing decides how to forward packets over the best path over the Internet based on feedback about "path metrics." In wireless networks, radio resource management is a closed-loop control that ensures that Wi-Fi devices use the best RF—connect to the least loaded channels, ones that are not interrupted by interference, or not crowded out by other devices.
Now closed-loop network systems are becoming more sophisticated and capable. It's harder to always deliver the perfect wireless network experience on a university campus or a 20-story office building than to maintain a backyard pool at 82 degrees and eight feet deep. In a highly dynamic system like Wi-Fi, it's not always apparent how to adjust for the equivalent of the perfect amount of water at the right temperature.
To date, the industry has basically given network operators a bunch of knobs to configure their networks. You need to be an expert to adjust those knobs for the perfect user experience. There's no autopilot. Yet what defines the perfect user experience may differ, depending on what services the user is consuming, what devices they are using, whether they are mobile or static, outdoors or indoors. Operators may have different objective functions that drive the network design. And how to achieve those objectives may also change over time based on environmental parameters.
What Does Automated Service Assurance Mean?
We're in the early days of automated service assurance for networks. Achieving this goal requires a network that can take higher-level business policy as input and convert it to a detailed configuration, and then generate and validate the resulting design and configuration.
The Wi-Fi design and configuration for classrooms, office buildings and outdoor spaces are unique for sure. But there are other optimizations, too. A design priority may be to optimize so that there is Wi-Fi coverage everywhere. Maybe the business goal is to minimize energy consumption. Or maybe it's critical to ensure everyone on the executive floor has a premium wireless experience.
The industry is innovating on creating these business abstractions that will enable a relatively unskilled operator to instruct the network to behave how he wants it to work. From there, network automation and orchestration tools are used to automatically configure the network. It's zero-touch provisioning on steroids, if you will.
A real-time awareness of the network state is critical to assure service. Today, networks are outfitted with sensory apparatus that can accurately measure the current state of the system. That lays the foundation to collect and analyze the sensory data, which can be used to compare the business policy to actual performance, and automatically make adjustments and take corrective actions.
Behind the scenes, tens of thousands of sensors may need to communicate back to a central point of control, where the telemetry data is stored, analyzed and acted upon. Advances in cloud, big data and machine learning make it possible to process these massive amounts of sensory data and to discern what's important enough to act on and what's irrelevant clutter. That simply wasn't possible a decade ago.
An Exciting Time in Networking
At Aruba, we're making great strides with automated service assurance. We're working closely with customers to capture the most pertinent scenarios, create the business policy abstractions, and then develop the tools that will allow operators to specify and execute their intent. We're leveraging our best practices for delivering great Wi-Fi—and going well beyond with big data analytics, machine learning, cloud, and engineering innovation. The end result is a network that does what you want it to—without the heavy lifting.
Related Blogs
Cars drive themselves. Planes have auto-pilots. Why do we still need people to run networks?
Would You Trust Changes to Your Network Recommended by a Machine?
