摘要
为了计算需要部署的应用实例数量和租用的虚拟数量,并建立租户与应用实例以及应用实例与虚拟机之间的放置关系,提出一种基于成本优化的多租户SaaS应用放置算法。通过分析多租户SaaS应用服务模式,提出了资源消耗度量模型和多租户SaaS应用优化放置问题的形式化描述;依据租约关系计算初始的应用实例数量和虚拟机数量,以此确定放置方案的编码方式,并采用基于虚拟机序列的遗传算法来选择最优的放置策略;通过实验验证了所提算法的可行性和有效性。
To calculate the amount of deployed SaaS application instances and leased virtual machines, and to establish the placement relationship between tenants and application instances as well as application instances and virtual machines, a multi-tenant SaaS application placement algorithm based on cost optimization was proposed. Through analyzing the service mode of multi-tenant SaaS applications, the resource consumption measurement model and the formal description of multi-tenant SaaS application placement problem were proposed. According to the lease relation- ships, the initial amount of application instances and virtual machines was calculated to determine the encoding of chromosome, and the best optimal placement strategy was selected by using greedy strategy-based genetic algo- rithn. The feasibility and effectiveness of the proposed algorithm were validated through experiments.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2014年第6期1508-1518,共11页
Computer Integrated Manufacturing Systems
基金
国家科技支撑计划资助项目(2013BA17F02)
山东省科技攻关资助项目(2011GGX10108
2010GGX10104
2010GGX10116
2010GZX20126200)
威海-哈工大共建项目(2011DXZJ07)~~
关键词
多租户
软件即服务
放置
虚拟机序列编码
遗传算法
multi-tenant
software as a service; placement; virtual machine order coding; genetic algorithms