基于STAGING与业务预测的服务部署优化与迁移方法

发布时间：2018-04-16 06:18

本文选题：多租户 + 优化部署　；参考：《山东大学》2014年硕士论文

【摘要】：传统的企业客户通常需要在软硬件、技术人员以及技术维护上花费大量的资金,然而,随着互联网技术的高速发展,他们通过支付一定的租赁服务费用,便可通过互联网享受到相应的硬件、软件和维护服务,以及享有软件使用权和不断的升级,不仅减少了运营成本,还提高了运营效率,这种最新的、最具效益的互联网运营模式被称为SaaS(软件即服务)。它是一种基于互联网提供服务的应用模式,服务提供商以许可的方式按需向客户提供应用,而提供服务的方式也是多样化的。 SaaS平台,即运营SaaS软件的平台,采用多租户模式,然而,对于多租户服务交付系统,租户要求服务提供商能满足其对服务的性能需求和其它QoS要求,比如,分配的硬件资源是否充足,服务响应时间是否符合租户需求等。而服务提供商则要求SaaS平台应该具备充分利用资源的能力。由此可见,租户希望租用的服务是高性能的,而提供商则要求在满足租户需求的同时最小化成本,但是,服务的高性能与低成本之间往往是矛盾的。因此,为了平衡两者的矛盾,本文采用了基于服务初次部署和运行时的两阶段优化部署策略。本文以已开发的供应商管理系统为基础,开发了基于该系统的benchmark(S-BM)。对租户应用初次的优化部署,文章采用了不断迭代基于S-BM的Staging测试过程的方法。而对于运行时的二次优化部署,采取了结合实时负载衡量模型和基于马尔科夫链的关键业务预测,并综合迁移成本制定服务迁移策略的解决方案。并且,文章对这两种方法进行了设计与实现,还通过实验分析了其有效性。此外,本文还总结与展望了所研究的工作。
[Abstract]:Traditional enterprise customers usually need to spend a lot of money on hardware and software , technicians and technical maintenance . However , with the high - speed development of Internet technology , they can enjoy the corresponding hardware , software and maintenance services through the Internet , as well as the right to use and upgrade of software , which not only reduces operating costs , but also increases operational efficiency . This latest and most effective Internet operation mode is referred to as SaaS ( software as a service ) . It is an application model based on the Internet to provide services , and the service provider provides applications to customers in a licensed manner , and the manner in which services are provided are also diversified .

SaaS platforms , i.e . platforms that operate SaaS software , employ multi - tenancy mode , however , for multi - tenant service delivery systems , tenants require service providers to meet their performance requirements and other QoS requirements for services , such as whether the allocated hardware resources are adequate , whether the service response times meet the tenant requirements , etc . The service provider requires that the SaaS platform should have the ability to take advantage of resources . Therefore , it is often contradictory between high performance and low cost of the service . Therefore , in order to balance the contradiction between the two , this article uses two - phase optimization deployment strategies based on service initial deployment and operation .

Based on the developed supplier management system , this paper develops a benchmark ( S - BM ) based on the system . For the first - time optimal deployment of the tenant , the paper adopts the method of iterative test process based on S - BM .

【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】