Server Busy Times Finder and Synthetic User Log Generator On-Demand Architecture

Abstract

As user data becomes more valuable, research in the field to simulate and generate the data using new ways is further advancing. Finding sufficiently many new and realistic user log data to load test the system with, is not always possible. In this research, a “system idle time finder” module which counts user logs to find the least busy times, and a “scenario generation module” which fits the probability density function to user logs and generates test scenarios have been prototyped. Prototypes have been integrated with the deep reinforcement learning-assisted crashless test module, and generated scenarios have been tested automatically at the least busy times.