Automating Computer System Configurations for Maximum Performance with Minimum Latency Variance

Computer system configurations with multiple tuning parameters can require time and effort to find the optimal settings. This research praxis aims to use load testing tools, the Universal Scaling Law, and optimization algorithms to reduce thetime and effort of finding optimal configuration settings. In this research, the author demonstrated the reduction of time and effort by 91% within 10% of global maximum for targeting the optimum configuration settings to maximize throughput and minimize response time latency variance. In addition, another more brute force meta-heuristic algorithm, the Hill-Climbing algorithm, is also demonstrated to have 100% accuracy in this research with an 81% reduction in time and effort. All data collected in the praxis was collected by running the target applications in Docker containers.

Author

• Newsom, James

Language

• en

Keyword

• Heuristic Machine Learning
• Performance
• Universal Scaling Law

Date created

• 2022

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Engineering Management

Degree

• D.Engr.

Advisor

• Etemadi, Amir

Committee Member(s)

• Kamp, John
• Widener, Donald

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/r207tq15z