XOR Hashing Algorithms to Measured Flows at the High-Speed Link

Abstract

Flow sampling algorithms and data streaming algorithms have widely adopted the hashing algorithms to detect flow information at the high-speed links for the applications of the flow distribution, the number of flow, the heavy-tailed flow, and so on. Hashing algorithms involve transforming a key inside a hash value. The hashing algorithms used in high-speed networks have some specific properties: uniformity of distribution, computation speed. In this paper, we analyze the XOR operator in the hash-based flow selection function at the high-speed links, and prove that bit XOR operator can increase the uniformity of distribution. The hashing algorithms on the XOR operate are evaluated by mathematical analysis and trace-driven experiments on CERNET.