Artists, Scholars, & Innovators Lecture Series

The Artists, Scholars, & Innovators Lecture Series, hosted by the Center for Teaching and Learning, is presented by award-winning faculty with artistic, scholarly, or innovative achievements. Please see below for information about upcoming presentations.

Fall 2020 Presentations

Automatic Detection of Performance Deviations in the Load Testing of Ultra-Large-Scale Systems

Presented by:

Dr. Haroon Malik

  • Assistant Professor, Department of Computer Sciences and Electrical Engineering
  • Winner of the 2019-20 MU Distinguished Artists & Scholars Award (Junior Recipient in All Fields)

Monday, October 12, 2020
4-5 pm | Virtual (live, remote)

About the lecture: Ultra-Large-Scale Systems (ULSS) — such as those developed by Google, Blackberry, Amazon, and worldwide banking systems — are service-oriented systems and generate revenue by providing composite services to a large user base. Maintainers of ULSSs strive to ensure that required service-level-agreements (SLAs) are met; failure to do so can result in huge monetary loss and costly lawsuits.

Maintainers/Analysts use load testing to detect early performance problems in a system before they become critical field problems. During the course of a load test, a system under test is closely monitored and large volume of performance counter data — usually, terabytes in size — is logged. The information helps them to observe the system’s behavior under load by comparing it against the documented behavior of a system or with an expected behavior.

In cases where such documented behavior does not exist, analysts often use a few of the performance counters as ‘rules of thumb’, observed from past practice or known to them from domain gurus, to gauge the performance of the system during a load test. Nevertheless, in ULSS evolving at a fast pace, ‘rules of thumb’ can be misleading at times. In the talk, I will present methodologies to help performance analysts to detect performance deviations, assist in the root-cause analysis of the detected deviations, and identify a special kind of anomaly called ‘discontinuity’.

 

Questions? Email ctl@marshall.edu.