Hypergraph-Based Data Reduced Scheduling Policy for Data-Intensive Workflow in Clouds

• 作者: Hui Kang Jia Li Jiao Yang Meiguang Zheng Xinxin Zhang Yong Tao Zhigang Hu
• 来源期刊: 国际计算机前沿大会会议论文集
• 学科: 社会科学
• 关键词: DATA-INTENSIVE WORKFLOW Directed HYPERGRAPH DATA REDUCED scheduling Cloud computing
• 年，卷（期）: 2017,（2）
• 页码范围: 80-82
• 页数: 3页
• 分类号: C5

摘要

Data-intensive computing is expected to be the next-generation IT computing paradigm. Data-intensive workflows in clouds are becoming more and more popular. How to schedule data-intensive workflow efficiently has become the key issue. In this paper, first, we build a directed hypergraph model for data-intensive workflow, since Hypergraphs can more accurately model communication volume and better represent asymmetric problems, and the cut metric of hypergraphs is well suited for minimizing the total volume of communication.Second, we propose a concept data supportive ability to help the presentation of data-intensive workflow application and provide the merge operation details considering the data supportive ability. Third, we present an optimized hypergraph multi-level partitioning algorithm. Finally we bring a data reduced scheduling policy HEFT-P for data-intensive workflow. Through simulation,we compare HEFT-P with three typical workflow scheduling policies.The results indicate that HEFT-P could obtain reduced data scheduling and reduce the makespan of executing data-intensive workflows