Quantifying the impact of network latency on the end-to-end response time of distributed applications

摘要

A method for measuring system response sensitivity, using live traffic and an analysis that converts randomly arriving stimuli and reactions to the stimuli to mean measures over chosen intervals, thereby creating periodically occurring samples that are processed. The system is perturbed in a chosen location of the system in a manner that is periodic with frequency p, and the system's response to arriving stimuli is measured at frequency p. The perturbation, illustratively, is with a square wave pattern.