‘Self-Driving Network’ is a term used to describe one facet of Software Defined Networking: When network conditions deteriorate, the network itself takes corrective action without human intervention (or with minimal operator supervision). My friend and Juniper CTO, Kireeti Kompella, eloquently describes self-driving networks as follows:
“When networks are self-driving, there will be automatic service placement and service motion, specific upgrades based on configured services, and inductive network response based on machine learning. Maintenance will be proactive rather than reactive (as it usually is today), meaning that components and systems will be fixed before an outage happens instead of the failure occurring and causing traffic disruption.”
An example of a self-driving network use case is changing traffic paths automatically to alleviate congestion caused by a link failure. Packet Design already demonstrated this in XL Axiata’s production network with our Explorer SDN Traffic Engineering App and using the OpenDaylight SDN controller for provisioning network changes.
Working with our SDN integration partner, Ciena Blue Planet, we recently demonstrated another self-driving network use case for a European mobile operator. This use case was much more sophisticated, and it was great to showcase the power of the two systems working together. As CRC errors started to occur on network links, long before the links eventually failed, our Explorer Suite (Performance Explorer module) notified the Blue Planet orchestration system which then checked for critical services running over these links and whether the policy for these services allowed proactive rerouting of traffic. If affirmative, Blue Planet instructed our Explorer SDN Path Provisioning App to compute new paths that avoided these troubled links in the network. The computed paths were passed to Blue Planet which configured the service head-end routers.
This is an important accomplishment because the service traffic was rerouted even before it experienced significant degradation – that is, proactively!
The above use cases illustrate the importance of network telemetry and analytics. Network performance data needs to be collected in real time, analyzed and baselined using machine learning, and alerts issued when performance deviates from the baseline. This is precisely what Performance Explorer does.
The real SDN breakthrough though, is the ability to intelligently automate remediation. A new path must be identified that has the capacity needed by the impacted services. To do this, traffic matrices must be calculated and any reservations for services in the network must be accounted for. The Explorer Suite’s Traffic Explorer module provides this data which is then used by our Explorer SDN Path Provisioning App to compute the new path(s).
The whole solution needs to be orchestrated and this is the job of the Blue Planet Orchestrator. It interacts with all components using open REST APIs and sets up the new paths in the network using resource adapters to commercial routers from various vendors.
One of the most expensive components of Google’s self-driving car is the LiDAR. The LiDAR collects real-time telemetry from the car’s surroundings and its high cost illustrates the importance Google places on telemetry. To use Kireeti’s analogy, self-driving networks need a LiDAR.
A self-driving network is in your future and the Explorer Suite can be your network’s LiDAR as well as its brain for making real-time path decisions.