%A QI Ping, WANG Fu-cheng, WANG Bi-qing, LIANG Chang-yong %T Dynamic level scheduling algorithm for cloud computing based on failure regularity-aware %0 Journal Article %D 2019 %J JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE) %R 10.6040/j.issn.1671-9352.3.2018.143 %P 103-115 %V 54 %N 1 %U {http://lxbwk.njournal.sdu.edu.cn/CN/abstract/article_3025.shtml} %8 %X Due to the characteristics of dynamic, heterogeneity and distributed, parallel tasks in cloud computing environment cannot be accomplished because they are vulnerable to resource node failure and communication link failure. Aiming at the problem that the reliability of dynamically providing cloud resources is low and the parameters of resource failure law are dynamically changing under the failure recovery mechanism. Firstly, the local characteristics of failure nodes and communication links in different periods are described by using Weibull distribution. Then, according to the analysis of various kinds of interaction between parallel tasks, a resource reliability evaluation model based on variable parameter failure rules is proposed. Finally, the model is incorporated into the DLS algorithm to obtain the reliable dynamic level scheduling algorithm CFR-DLS, so that the reliability of alternative resources is fully considered when calculating the scheduling level. Simulation results show that the proposed CFR-DLS algorithm can greatly improve the reliability of cloud services while only increasing a small amount of additional failure recovery overhead.