Performance Comparison for Distributed Computing Model in Long-Running Financial Analysis Computation

• Main Authors: JYUN SHENG HE, 何峻昇
• Other Authors: Kevin Chihcheng Hsu
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/72053177491821401945

碩士 === 國立中央大學 === 資訊管理學系 === 104 === Currently the global financial field, there are many studies about using existed financial analysis model as applications, but there are some technical problems. For example: to calculate complex financial analysis models and handle big data takes lots of computing time. for the current CPU Intensive and I/O Intensive problem, research distributed computing literature no fairer standard and development processes, to make the developers can’t know what kind of distributed computing for solving Intensive what type of properties, and how to through a cost-effective manner to distributed computing. In this study, for the financial field faced CPU Intensive and I / O Intensive problem, using distributed computing model to enhance the financial analysis model and big data on the efficiency of operation, thereby solve lot of computing time, and comparison of three common distributed computing model, in the CPU Intensive and I/O Intensive characteristic operation efficiency, to make developers can choose more suitable for the characteristics of the distributed computing model development, enhance the efficiency of financial models in operation, but the result is not the same as the original expected results, research how to amend the original process to quickly identify the cause of inefficiency. The study presents a revised preliminary process can really quickly identify poor efficiency reasons. It divided into seven phases, the first step, according to CPU Intensive and I / O Intensive properties, selected more suitable for financial calculation model, the second step, to determine the programming language more suited to the development of complex operations, the study is the use SAS and MATLAB programming language development, the third step, for the program to be converted, to compare the initial efficiency, the fourth step, to ensure that the program did not increase the complexity of the case, to make SAS program convert MATLAB program, the fifth step, develop distributed computing program, the sixth step, experiment, finally, efficiency analysis and discussion of distributed computing model.