There is an increasingly focus on Programmable Systems for controlling edge networked systems. During the normal operation of these systems a high number of unpredicted new flows can be suddenly presented to the system control level, requesting novel flow rules to successfully steer the associated data toward their final destinations. The extra control required by these new flows can increase the channel control workload up to a level, the available active controllers are not able to process that workload in a satisfactory degree. The current work investigates a distributed solution, which online acquires the control channel workload and adjusts the number of active controllers to variations on that measured workload. The controller scaling provides an elastic and efficient control plane for dynamic edge programmable systems. Thus, our work discusses the design, deployment and evaluation of the proposed framework. The obtained experimental results validate the main objectives of the multicast-based decentralized and scalable control orchestration, which positively accommodates the unexpected changes on the number of new data flows traversing the controlled edge network infrastructure, using only minimal information shared among controllers and diminishing the control channel workload.

Jose Moura acknowledges the support given by FCT / MCTES through National Funds When Applicable Co-Funded European Union (EU) Funds under Project UIDB/50008/2020 and in part by Instituto de Telecomunicações.

Decentralized,Control,Workload,Adaptive,Orchestration,Edge networked system