End-to-end Network Throughput Optimization Through Last-mile Diversity

In this paper we propose a platform that optimizes the available end-to-end throughput in real time through overlay networks. With the knowledge the network topology and conditions, it strives to achieve the optimal end-to-end throughput by exploring the last-mile diversity. It allows the flexible and responsive per-end-user selection of the edge node for the overlay networks, and thus can fast recover from network failures or performance degradation. We present our design of the end-toend throughput optimization system with detailed discussion of each component including dynamic routing engine, performance monitor and information exchange. Our experimental results from a real-world deployment show that it not only brings up to 5 times throughput gains in the presence of 0.05% loss, but also improves 20% throughput when facing delay increase in the original path.