How to Estimate Web Application Concurrent Events Using Either Load Generator Data or Queueing Model Results

EasyChair Preprint 6029, version 2

Abstract

Some of the most sought-after web application load testing results are the concurrent GET/POST event quantities used to dimension internal system resources such as process thread pools and input queue buffers. The concurrency information desired is the proportion of time there are 0, 1, 2, etc. web events resident in the system under test. These proportions are not always available from target system measurements but, as will be shown, can be approximated with data produced by the load generator, or estimated using queueing models. The data analysis and queueing model tools that support the ideas presented were developed by this author and contained in two of his GitHub repositories, web-generator-toolkit2 and QueState, as free open source downloads.

Keyphrases: Computer performance, Load Generator, Queueing models, State Probabilities, statistics, user concurrency

@booklet{EasyChair:6029,
  author    = {James Brady},
  title     = {How to Estimate Web Application Concurrent Events Using Either Load Generator Data or Queueing Model Results},
  howpublished = {EasyChair Preprint 6029},
  year      = {EasyChair, 2021}}