Evaluating the performance of multi-path routing and congestion control in presence of network resource management

Network traffic is increasing in size and is becoming more and more dynamic leading to unpredictable and highly variable traffic patters. Multi-path routing is a valuable online technique to deal with such trend, mostly for intra-domain TE, multi-homing, wireless mesh networks, metropolitan access networks and has been shown efficient for a large spectrum of future traffic scenarios. In this paper we present MIRTO, a distributed multi-path routing protocol that jointly uses best path selection and flow control for optimality and stability. MIRTO is feedback based and is designed around some key observations inferred from optimization. We analyze MIRTO's performance and we compare it with TEXCP and TRUMP, two other recently proposed multi-path routing algorithms. On a US-like backbone network, with and without in-network fair queuing schedulers, our algorithm proves to work as better as the other two while relying on simpler feedbacks and consuming less network resources. Modeling and analysis of such algorithms is performed through fluid models, based on ordinary differential equations (ODEs).