SaaS环境下基于服务质量的资源分配算法  

Quality of service(QoS) based resource allocating algorithm in software as service(SaaS)environment

作　　者：张志超[1] 彭蓉[1] 黄华[1,2] 

机构地区：[1]武汉大学计算机学院软件工程国家重点实验室,湖北武汉430072 [2]景德镇陶瓷学院信息工程学院,江西景德333001

出　　处：《计算机工程与设计》2015年第8期2090-2097,2102,共9页Computer Engineering and Design

基　　金：国家自然科学基金项目(61170026);武汉市科技攻关计划基金项目(201212521826)

摘　　要：为保证SaaS环境下租户的服务质量,最小化SaaS提供商的底层资源成本,提出基于服务质量的资源分配算法。采用多租户共用虚拟机的方式节省底层资源,利用协调虚拟机来处理增长的租户请求,研究SaaS环境下租户请求的变化规律,确定协调虚拟机的数量。根据协调虚拟机的使用状态,实时提取适量资源,适应租户请求增长;在租户请求量减少时,采用慢收缩策略,防止租户请求量反弹。模拟实验结果表明,该算法可以有效减少服务级别协议(SLA)违背,最小化底层资源成本。To guarantee QoS for tenants in SaaS environment and minimize the infrastructure cost for SaaS providers, OoS based resource allocating algorithm was proposed. The mean of sharing virtual machines among tenants was adopted to save infrastruc- ture resources. The variation rules of tenant requests in SaaS environment were researched to determine the amount of coordina- ting virtual machines used to solve the increasing requests. Proper resources were provisioned in time according to the using state of the coordinating virtual machines when tenant requests increased. Slow scaling down policy was adopted to prevent tenant re- quests rebound. Simulation results demonstrates that, the proposed algorithm can effectively reduce SLA violations and minimize the cost of infrastructure resources.

关 键 词：软件即服务 多租户 服务级别协议 资源分配 服务质量 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]