A first look at inter-data center traffic characteristics via Yahoo! datasets

摘要: Effectively managing multiple data centers and their traffic dynamics pose many challenges to operators, as little is known about the characteristics of inter-data center (D2D) traffic. In this paper we present a first study D2D using anonymized NetFlow datasets collected at border routers five major Yahoo! centers. Our contributions are mainly two-fold: i) develop novel heuristics infer IP addresses localize locations from datasets, ii) analyze both client correlations between these two types reveals that uses hierarchical way deploying centers, with several satellite distributed in other countries backbone US locations. For separate client-triggered background aggregate port based correlation, respective characteristics. findings shed light on interplay within large content provider, provide insights designers operators well researchers.