Resilient Disaggregated Network Flow Monitoring
Yongquan Fu, Dongsheng Li, Siqi Shen, Yiming Zhang, Kai Chen · 2019
Sketch has been extensively used for scalable network monitoring, which unfortunately, is sensitive to hash collisions. Deploying the sketch involves fine-grained performance control and instrumentation. This paper presents a new class of sketch structure that proactively minimizes the estimation error and reduces the error variance. We develop a disaggregated monitoring application that natively scales the sketching deployment. Testbed and real-world trace-driven simulations show that LSS achieves close-to-optimal performance under hash collisions.