Wenji Wu, et. al., G-NetMon: A GPU-accelerated Network Performance Monitoring System, here. Seems like something I would like for SOR and hooked up to Corvil.
The main purpose of this work is to explore new opportunities in network traffic monitoring and analysis with GPUs. The experiment results show that GPU can significantly accelerate the flow data processing, by a factor of 5.38 (G- NetMon vs. H-CPU), or by a factor of 33.08 (NH-S-NC-GPU vs. NH-CPU). At present, G-NetMon is designed to detect sub- optimal bulk data movements. In the future, we will enhance it with security features. To implement security features, G- NetMon needs to examine flow records with security rules one by one in real time or semi-real time, which require more computation capabilities. The computation pattern of examining flow records with security rules one by one is similar to that of the non-hash-table implementations discussed in the paper, in which GPU can significantly accelerate the flow data processing.