公有云计算服务组合系统收益优化研究

项目名称： 公有云计算服务组合系统收益优化研究

项目编号： No.61202040

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 鲍亮

作者单位： 西安电子科技大学

项目金额： 25万元

中文摘要： 公有云计算服务组合系统是"软件即服务"模式下的一种新型应用系统，在目前的在线服务中得到了广泛的重视与应用。收益优化是公有云计算服务组合系统的重要决策与目标。现有研究均以组合服务请求固定不变为假设前提，且很少考虑系统的收益优化问题。在实际环境中，用户请求不断变化，只能得到模糊的请求比例信息。除了请求信息短期随机波动外，系统还要面对组合服务调整等长周期稳定性变化，这时需要进行再平衡。本项目以需求模式动态变化为切入点，研究模糊组合服务请求比例信息下公有云计算服务组合系统的收益优化与再平衡问题。通过分析请求比例动态变化对服务调整、服务分配等因素的影响，明确模糊请求比例下系统收益优化与再平衡问题的特征，建立问题的优化模型，设计求解算法，并通过实践加以检验和修正。最终有望为需求模式动态变化环境中的公有云计算服务组合系统的优化设计提供新理论与方法，为提升系统的管理水平与服务效率提供理论指导。

中文关键词： 公有云服务组合系统；收益优化；正交试验；遗传算法；用户行为分析

英文摘要： Service composition systems in public cloud computing environment (public cloud service composition systems for short) are prevalent and widely deployed as online services with the boom of "Software as a Service". Previous researches on public cloud service composition system suffer from two limitations: 1) these researches are often based on an idealized assumption of fixed requsts from clients which is not always reasonable in practice, and 2) they only emphasize on providing high quality of service for clients and lack of the consideration of profit optimization for service providers. In a practical circumstance, a public cloud service composition system confronts the ongoing-changed requests from a huge number of clients and hence gets fuzzy request information; besides, the long-term bussiness adjustments, which will lead to system rebalance, should also be concerned. This project studies the profit optimization and rebalance problem of public cloud service composition system with the fuzzy request information. The core features of this problem needs to be clarified firstly by analyzing the influences of service adjustment and assignment from the fuzzy requst information. After that, some optimization models and algorithims should be designed, tested and adjusted with exprimental and practical data. Some ne

英文关键词： public cloud service composition system；profit optimization；orthogonal design；genetic algorithm；user behavior analysis