Queuing Network Models on Open Systems

by Michael Morse

Submitted : Spring 2009

This project covers the methods used to maximize transactional throughput (TPS) in a highly scalable and resource intensive environment. This will be done by running analysis on each of the individual components (CPU, memory and disk) and identifying the processing bottlenecks. By doing this we determine the current TPS break point, and what cost effective upgrades can be made to cover anticipated growth. The data is all determined and arranged through the use of queuing network models (QNM). QNM requires the complete understanding of work (processes running on a system) and individual hardware components. The QNM utilizes the relationship between work and hardware to support estimates of throughput, response time and queue lengths.

[ Back ]