A Distributed Report Generating Approach by Effectively Utilizing Enterprise Computing Power

• Main Authors: Tsung-Hsien Yang, 楊宗憲
• Other Authors: Da-Yin Liao
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/77136903815838343273

碩士 === 國立暨南國際大學 === 資訊管理學系 === 94 === Reporting systems are indispensable to enterprise’s daily operations. People rely on reports to understand the business status quo as well as to make their decisions. The process of generating reports is usually data-intensive and thus tedious. This thesis deals with reports generating problems in an enterprise. As there are precedence relationships among the input and output data of reports, generating reports must follow the order of reports. We develop an agent-based, distributed reports generating framework that can effectively utilize enterprise’s computing power, especially for the lower-utilized personal computers (PCs) on clerks’ desktops. In the framework, each report generating job is done by a distributed report generating agent (RGA). The collaboration and coordination of reports generating jobs are executed by a report coordinating agent (RCA) which is responsible to schedule report generating jobs to distributed RGAs. We adopt a heuristic approach to solve the distributed reports scheduling problem, which is NP-hard in computational complexity. Our report scheduling method schedules report generating jobs in a greedy way such that the overall report generating time is minimized. We implement a distributed reports generating test environment with five personal computers. Some experiments are designed and conducted to validate our idea on the agent-based, distributed report generating framework. Numerical results indicate that our approach is feasible and effective. We conduct experiments to explore the computation times and the reports generating makespan of the proposed reports scheduling method, as compared to the optimal solutions obtained from exhaustive search. We find that our reports generating scheduling algorithm performs well in the scenarios whose capabilities of computers are almost equal to each other.