What is the Self-Driving Network?

The Self-Driving Network™, similar to a self-driving car, is an autonomous network that is predictive and adaptive to its environment. It simultaneously increases economies of scale and efficiencies, while decreasing operating costs. As a result, it delivers an optimized and customized quality of experience inexpensively to the end-user. When automation frameworks are infused with telemetry, big data analytics, and machine learning, and network guidance is provided to self-analyze, self-discover, self-configure, and self-correct, the autonomous network is born. The aim of the Self-Driving Network is to eliminate the “manual work” required to keep networks running.

what is self driving network


A primary cost for telecommunications service providers and enterprises is operating their networks. It costs more money to maintain, monitor, and operate an existing network than it does to initially buy and install the networking equipment. By implementing a Self-Driving Network that encompasses all of the advantages provided by automation, companies can redeploy their resources to focus on other tasks that provide a higher return on investment (ROI). This greatly benefits the individual (who no longer needs to devote time, energy, and labor on repetitive tasks), and the company as a whole (by having resources/capacity available to refocus their talent towards strategic activities requiring nuanced thought and reasoning).

what is self driving network
what is self driving network
Path to a Zero Touch Network

The path to the Self-Driving Network (zero touch) begins with Juniper Network’s three-step automation strategy:

1. Reduce operation complexity by simplifying and abstracting networks.

2. Enable customers to deploy new network services faster.

3. Improve capacity utilization and network resiliency through deep telemetry.

The route forward toward an autonomous network relies on telemetry, automation, machine learning, and programming with declarative intent:

  • Telemetry: We need telemetry based on push semantics and anomaly detection based on machine learning. Juniper’s OpenNTI is an example of a simple, open-source tool for collecting, normalizing, and visualizing key performance indicators (KPIs) using standard telemetry, analytics, and a hierarchical design.

  • Automation: We currently automate topology discovery, path computation, and path installation. We need automatic service placement and service motion, specific upgrades based on configured services, and inductive network response based on machine learning.

  • Machine Learning: Machine learning employs creative methods of programming by moving away from static coding towards dynamic algorithms that learn from data inputs, make predictions, and take appropriate actions. Juniper’s AppFormix solution combines the power of machine learning and streaming analytics with application awareness of orchestration systems such as Openstack and Kubernetes-based Hybrid clouds and NFV/Telco clouds.

  • Declarative Intent: Tell the network what results you want, not exactly how to accomplish it. Juniper’s Northstar tool enables service providers to install network paths based on provided constraints such as bandwidth, diversity, and inter-virtual network policies.

At first glance, this bold and compelling vision of the future of networking may seem ambiguous to many, impossible to some, and even fearful to a few. However, it’s exactly the type of revolutionary foresight that’s required as the industry manages disruptions influenced by: economics, technology, and society. A future world in which technology further disappears into the fabric of our lives makes us more productive, healthier, and freer to do more of what makes us happy. By following the path to the Self-Driving Network, we can realize that future.