Division And Parameterization Based Performance Tuning For Queue Network Model System

With the development of computer technology, performance is becoming essential to the success of a computer system, and it is one of the hot topics in the area of computer science. Many researchers and engineers have been exploring solutions to the performance of computer system, and various performance improvement methods have been studied and applied in the systems.As an important part of software performance engineering, software performance tuning plays a key role in improving the performance of computer systems. Basically, performance tuning consists of two areas: upgrading the hardware and improving the software performance. This thesis focuses on the area of improving the computer system performance by enhancing the software efficiency. Based on the existing studies and theories of software performance tuning, it explores a common process to tune the performance of queue network model like system, in which division and parameterization are used as the key means.To evaluate the effectiveness and applicability, the performance tuning process is applied to an actual queue network model based system: financial messages delivery system. As the business grows, this system is faced with more and more performance issues, which need to be solved so as to support the continuously growing business. Within the process of performance tuning, the division strategy is used to improve the whole system performance step by step and from bottom to top. And during the performance testing, the parameterization method is used to define the system performance which could be adjusted and combined to find out the best parameter configuration that ensures the best system performance. Finally, with a thorough result analysis and validation, this thesis brings forth feasible solutions to improve the system performance and to achieve the performance objectives.