异构云计算中心IaaS（Infrastructure as a Service）服务性能分析是IaaS云服务供应商必须要解决的问题。理论建模是行之有效的性能分析方法，但传统计算系统的模型成果没有考虑云计算中心规模庞大、参数多以及虚拟化等特征；而现有云服务模型成果或者对系统做了不切实际的假设，或者近似精确度较低，或者没有进行理论分析(包括整体模型的遍历性、算法收敛性等)。本项目旨在利用嵌入马尔可夫链和层次建模等技术，解决现有云服务模型成果存在的不足，建立能分析实际异构云计算中心IaaS服务性能的高精度且高效的模型，以此为基础来研究各种性能指标的计算方法，从而定量分析各种因素对云服务性能以及云中心成本的影响，促进异构云计算中心资源分配体系的完善。

Performance analysis of IaaS（Infrastructure as a Service）services in heterogeneous Cloud Computing Centers (CCC) is a critical issue to cloud service providers (CSP). One of the most widely used assessment techniques is analytical modeling. However, the existing approximate analytical methods for modeling traditional data centers did not consider the unique features of CCCs, such as large scale, large parameter number and virtualization. The existing modeling approaches to cloud services either made unrealistic assumptions of CCCs, or did not capture the cloud behaviors accurately, or did not carried out theoretical analysis, such as monolithic model’s ergodicity and algorithm convergence...This project aims to address the deficiencies in the existing modeling approaches to CCCs. We develop efficient and accurate models for heterogeneous large-scale CCC IaaS services by applying the techniques including embedded Markov Chain, hierarchical modeling, and so on. From these developed models, performance metrics for IaaS clouds are defined and the computing methods are developed. The outcome of this project is deemed to be beneficial to quantitatively analyzing various factors’ effects on cloud performance and CCC costs, therefore enhancing the heterogeneous CCC resource management system.