Multiple DAGs reliability model and fault-tolerant scheduling algorithm in cloud computing system

Weipeng Jing^{1, 2}, Yaqiu Liu^{1, 2}

¹The College of Information and Computer Engineering, Northeast Forestry University, Harbin 150040, China

²Heilongjiang province engineering technology research centre for forestry ecological big data storage and high performance (cloud) computing, Harbin 150040, China

 In this paper, in order to provide the reliable scientific workflow scheduling problem for cloud computing, a dynamic of RANK-Hierarchical algorithm is put forward which taking account of communication contention as well as supporting task dependencies (CCRH). A communication contention model is first defined, as soon as the earliest completion of the primary and backup task is deduced, besides the executive processor is limited, use dynamic hierarchical method and calculate of each DAG unfair degree factor for multiple DAGs scientific workflow. It can deal with the problem that multiple DAGs workflow comes at different time and have various kinds of structure. Both the theory and experiments have proved the algorithm not only improve the scheduling fairness of multiple DAGs workflow but also shorten the average execution Makespan effectively while meeting reliability constraints and meanwhile the produce well robustness.