Research on Weighted Rotation Fair Scheduling Algorithm Based on Hama Parallel Computing Framework
With the development of big data technology in the industry and academia,a large number of distributed computing platforms are put into operation,of which the Apache Hadoop platform is the most widely used.Hadoop shields the underlying implementation details of a distributed system,making application developers more focused on algorithmic logic.However,Hadoop has its limitations.Its efficiency is not high when dealing with problems such as graph computing and machine learning.The emergence of the Apache Hama parallel computing framework based on the BSP(Bulk Synchronous Parallel)model makes up for this shortcoming of Hadoop.However,because the time of Hama creation is not long and the time of development is short,it is still necessary to continue to improve in many places,especially its core module—job scheduler.The Hama currently used FCFS(first come first service)job scheduling algorithm,and it is not only unable to efficiently implement multi-user shared clusters,but also has a great impact on resource utilization of clusters.
Zhijie LIN Haifeng HUANG
国际会议
上海
英文
101-109
2017-12-30（万方平台首次上网日期，不代表论文的发表时间）