面向大数据试验场应用的资源优化分配

发布时间：2018-06-12 22:19

本文选题：大数据试验场 + 资源分配　；参考：《浙江大学学报(工学版)》2017年06期

【摘要】：针对多个用户申请有限的资源的情况,提出两段式优化资源分配方法,在满足用户需求的同时确保资源的公平有效分配.根据以往的执行日志自动确定单类应用相对于资源投入量的饱和点.根据占优资源公平分享原则,为每类应用动态确定可投入运行的实例数量,实现全局最优化配置,在平台层面体现资源共享、杜绝欺诈、公平最优以及帕累托均衡等特性.选择Clik+公用的基准测试程序集,以Docker容器作为应用运行环境验证应用的普遍性,实验结果表明:两段式优化资源分配方法提升了资源利用率,且在多用户同时申请资源时优化了资源配置.模拟并检验了两阶段优化资源分配算法的可行性及有效性.
[Abstract]:Aiming at the situation of multiple users applying for limited resources, a two-segment optimization resource allocation method is proposed to ensure the fair and effective allocation of resources while satisfying the needs of users. According to the previous execution log, the saturation point of single class application relative to resource input load is automatically determined. According to the principle of equitable sharing of dominant resources, this paper determines the number of instances that can be put into operation dynamically for each type of application, realizes the global optimal configuration, and embodies the characteristics of resource sharing, eliminating fraud, fair optimization and Pareto equilibrium at the platform level. The common benchmark assembly of Clik is selected, and Docker container is used as the application environment to verify the universality of the application. The experimental results show that the two-segment optimization resource allocation method improves the utilization of resources. Resource allocation is optimized when multiple users apply for resources at the same time. The feasibility and effectiveness of the two-stage optimal resource allocation algorithm are simulated and tested.
【作者单位】：复旦大学计算机科学技术学院上海数据科学重点实验室;
【基金】：国家自然科学基金资助项目(60873115) 教育部-中国移动科研基金资助项目(MCM20123011) 上海市科技发展基金资助项目(13dz2260200,13511504300)
【分类号】：TP311.13

【相似文献】