Research And Design Of Data Collecting And Analyzing Subsystem Of Lsf

With the development of Cluster technology, Cluster Job Management System becomes more and more popular in high-performance computer field. In view of LSF developed by Platform Computing Company, which is one of the Cluster Job Management System, this dissertation studies resource information data and job running data of the system; and it is analyzed that system administrator cannot make efficient and reliable decision based on inquiring and displaying of the current system data. Therefore, an assistant system is essential for administrator to inquire and analyze data conveniently.This dissertation makes a research on principal and job execution process of LSF system, analyzes two kinds of data and the current method of inquiring and analyzing data in LSF system, and concludes that data provided by system cannot satisfy the administrator's decision demands. So a Data Collecting and Analyzing Subsystem is designed, which devided into Data Collecting module and Data Analyzing and Transforming module. In design of data collection process, JNI technology is adopted to collect the data of system resource information; in design of data input process, data-lost protection mechanism is set up; in design of data analyzing process, ETL technology and Data Warehouse technology are used. Finally, general system functional test and performance test of modules are made, which confirm that subsystem is achieved the advanced design objective.