High-speed TCP flow record extraction using GPUs

Abstract

Traffic analysis is an essential part of capacity planning, quality of service assurance, and reinforcement of security in current telecommunication networks. Traffic volume increases with network speed and the analysis of large traffic traces is computationally intensive. The paper presents, for the first time ever, a flow extraction software that allows to obtain complex TCP-aware flow records at 4.4 millions of packets per second in a single GPU. Such TCP flow records include number of retransmissions and duplicates per flow, whose calculation is very challenging to obtain at high-speed. Our software significantly increases the processing performance of the recently proposed high-speed sniffers based on commodity hardware and demonstrates the advantages of applying massively parallel processing devices for traffic analysis.