A routing optimization algorithm of equal-cost-multi-paths based on Link Criticality

To effectively avoid congestion and take full advantage of bandwidth inside a network, routing optimization is needed for path selection over a set of equal-cost-multi-paths. In this paper, a Link-Criticality-based ECMP routing optimization algorithm is proposed. As the traffic in the network overlaps on some links, LCER selects the path with least link average utilization and hops for the ECMPs. As compared with the existing traffic splitting algorithms, LCER provides a network-wide load balancing performance, no packets arrived out-of-order, and the lowest average end-to-end packet delays. Although additional expected bandwidth, based on measured daily traffic profiles or service-level agreements, is needed by LCER, we argued that the extra complexity incurred is marginal.