Parallel routing on multi-core routers for big data transfers

Abstract

Over the last several years, the deployment of multi-core routers has grown rapidly. However, big data transfers are not leveraging the powerful multi-core routers to the extent possible, particularly in the key function of routing. Our main goal is to find a way to use these cores more effectively and efficiently in routing the big data transfers. We propose a novel approach to parallelize data transfers by using each core in the routers to calculate a separate shortest path. For each core, we generate a different "substrate" topology in order to allow shortest path calculations to find a different end-to-end (e2e) path. By abstracting a different topology for each core, we indirectly steer each core to calculate a different e2e path in parallel to each other. The e2e big data transfers can use these shortest paths obtained from each substrate topology to increase the total throughput. We present an initial evaluation of the concept.