CAJ | 학술논문

信息社会是数据爆炸的时代，分布式并行计算在应对海量数据处理等大作业方面非常有效．Apache基金会下的Hadoop正是这样一个开源的分布式系统基础构架．首先介绍了Hadoop平台下作业的分布式运行机制，然后对Hadoop平台自带的3种任务调度器做分析和比较，最后在分析JobTracker类文件的基础上指出了创建自定义任务调度器所需完成的工作．
신식사회시수거폭작적시대，분포식병행계산재응대해량수거처리등대작업방면비상유효．Apache기금회하적Hadoop정시저양일개개원적분포식계통기출구가．수선개소료Hadoop평태하작업적분포식운행궤제，연후대Hadoop평태자대적3충임무조도기주분석화비교，최후재분석JobTracker류문건적기출상지출료창건자정의임무조도기소수완성적공작．
Distributed parallel computing has been proved a more effective way in dealing with mass data. The Hadoop platform developed by Apache Foundation is just such an open source distributed infrastructure. This paper firstly intro- duced the distributed operating mechanism of task on Hadoop. Secondly, it released 3 task dispatcher on Hadoop and did the analysis and comparison based on them. At last, on the analysis of JobTracker class file, this paper pointed out how to cre- ate a self - defining task dispatcher.