Visual Analysis of Distributed Search Traffic in a Peer-to-peer Network

We study distributed searches in fully decentralized peer-to-peer networks. This papers reports on a study that aims to understand the robustness and load balancing of a distributed search network. We focus on a visual analysis of search traffic, load distribution, and edge density when the network experiences adversaries. The analysis reveals that the network tends to rely more heavily on an even smaller portion of nodes for searches when an increasing number of nodes become unavailable. In the meanwhile, nodes become more exploratory and tend to engage more of those previously unutilized. The trade-off between a more skewed traffic distribution and a greater edge density in related scenarios is important to the robustness of the network and requires further investigation.