Scheduling Multiple Workflows on HPC Cloud

• Main Authors: Jiang, Hejhan, 江和展
• Other Authors: Huang, Kuochan
• Format: Others
• Language: en_US
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/58721443346465938298

碩士 === 國立臺中教育大學 === 資訊科學系 === 99 === Cloud computing is getting popular in recent years. It provides several kinds of services for various users. High Performance Computing (HPC) cloud has recently become one of the promising cloud services. It provides on-demand high-performance computing services for compute-intensive scientific and engineering applications. Many large-scale applications are usually constructed as workflows due to large amounts of interrelated computation and communication. Most previous researches focus on single workflow scheduling. Since cloud has to serve many users simultaneously, how to schedule multiple workflows efficiently becomes an important issue in HPC cloud environments. Traditionally, list scheduling and clustering are the two most important workflow scheduling strategies. In this thesis, we propose a hybrid approach for multi-workflow scheduling, which takes advantage of both listing scheduling and clustering. For task allocation, we developed a distributed gap search scheme which outperforms existing approaches. The proposed approaches have been evaluated with a series of simulation experiments and compared to existing methods in the literature. The results indicate that our hybrid approach outperforms typical listing scheduling and clustering methods significantly in terms of average makespan, up to 12% performance improvement.