An Innovative MapReduce-Based Approach of Dijkstra's Algorithm for SDN Routing in Hybrid Cloud, Edge and IoT Scenarios

Nowadays, with the advent of Cloud/Edge Computing and Internet of Things (IoT) technologies, we are facing with a tremendous increase of network connections required by different new cutting-edge distributed applications spread over a wide geographical area. Specifically, the proliferation of IoT devices used by such applications and associated data streams require a highly dynamic network ecosystem; the traditional network technologies are not adequate to efficiently support them in terms of routing strategies. In order to deploy such applications, providers need an advanced awareness of the Cloud/Edge and IoT networks in terms of flexible packets routing that can compute the paths according to different parameters including, e.g., hops, latency, and energy efficiency policies. In this context, Software Defined Networking (SDN) has emerged as the answer to these needs decoupling control and data planes, using a logically centralized controller able to manage the underlying networking resources. In this paper, we focus on the adoption of Dijkstra's algorithm in SDN environments to support applications deployed in Cloud/Edge and IoT scenarios. Specifically, considering a highly scalable network topology that includes thousands of network devices, in order to reduce the path computation, we propose a revised MapReduce approach of Dijkstra's algorithm. Experiments show that, compared to the sequential implementation, the MapReduce approach drastically reduces the shortest path computation performance when considering a complex Cloud/Edge and IoT network topology including thousands of virtual network devices.