Classification of Load Balancing in the Internet

Resumo

Recent advances in programmable data plans, software-defined networks and the adoption of IPv6 support new and more complex load balancing strategies. In this work, we introduce the Multipath Classification Algorithm (MCA), a probing algorithm that extends traceroute to identify and classify load balancing on Internet routes. We generalize the current formalism to describe load balancing and extend existing measurement techniques to consider that load balancing can use arbitrary combinations of packet header fields. We propose optimizations to reduce the probing cost, applicable to both MCA and existing load balancing measurement techniques. Through large-scale measurement campaigns, we characterize and study the evolution of load balancing on the IPv4 and IPv6 Internet, using various transport protocols. Our results show that load balancing is more prevalent and that load balancing strategies currently in use are more mature than the previous characterizations have found. We share our measurement and analysis software as well as datasets with the Internet measurement community.