一种多租户SaaS应用部署优化算法  

An Optimization Algorithm for Multi-tenant SaaS Applications Deployment

作　　者：曹祖凤[1] 孟凡超[1] 周学权[1] 初佃辉[1] 

机构地区：[1]哈尔滨工业大学(威海)计算机科学与技术学院,山东威海264209

出　　处：《计算机工程》2013年第10期14-18,23,共6页Computer Engineering

基　　金：国家科技支撑计划基金资助项目(2013BAH17F02);山东省科技发展计划基金资助项目(2011GGX10108;2010GGX10104;2010GGX10116;2010GZX20126200);威海市科技发展计划基金资助项目(2011DXZJ07);哈尔滨工业大学(威海)校科学研究基金资助项目(HIT(WH)XB200901)

摘　　要：在软件即服务(SaaS)平台构建初期,存在仅已知客户的需求和服务器类型,而所需服务器/虚拟机和SaaS应用实例数量不确定的情况。为此,提出一种多租户SaaS应用部署策略的优化算法。设计多租户资源消耗模型和优化部署问题模型,采用基于贪心策略的遗传算法对模型进行求解,给出优化的部署策略。实验结果表明,该算法具有较好的适应性和扩展性,给出的策略能为平台构建的实际部署提供理论依据。At the beginning of Software as a Service（SaaS） platform construction, service providers need to invest the initial cost of the software and hardware environment, rente server and deploy the application instances. As only customers＇ needs and the types of server are known for the initial deployment, while the amount of server/virtual machines and SaaS applications needed is unknown. To solve optimization deployment problem under that uncertainty, this paper puts forward a multi-tenant SaaS optimization deployment model, including multi-tenant resource consumption model and optimization deployment problem model, uses a Genetic Algorithm（GA） based on greedy strategy to solve the model, and gives optimized deployment stratege. Experimental result shows that this algorithm has better adaptability and scalability. The given strategy can provide a theoretical basis for the actual deployment of platform constructiong.

关 键 词：多租户 软件即服务 优化部署 资源消耗模型 贪心策略 遗传算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]