Performance Analysis And Optimization Research Of IaaS Cloud Service

Cloud computing is a common research focus in the field of both current academic and industrial circles, in this era of cloud computing, users get access to services from cloud service providers, the Quality of Service directly affects users’ satisfaction, therefore the Quality of Service is vital for large-scale commercial applications with cloud services system. Infrastructure-as-a-Service cloud is the foundation of cloud services model whose performance can directly affect the whole performance of a cloud service system. It is necessary to study and analyze the performance of the Iaa S cloud service because it can be used to evaluate the performance of a cloud service system and acted as a reference for future performance optimization of a system. In this paper, we analyze the basic model of Iaa S cloud services platform, the system model with processor failure and repair and the parallel processing system model in detail respectively. Not only the failure and recovery mechanism of service computing node are considered in cloud service system model presented in this study, but also the load balancing when service node conduct tasks in parallel is considered.In view of the need to accurately assess and analyze the Performance of Iaa S cloud to ensure the Quality of Service and avoid violation the Service Level Agreements requirements, an analysis model based on Queuing Network is proposed to analyze a process cloud service requests are handled. An analysis model based on Queuing Network is proposed to process cloud service requests. Through the analysis of underlying Markov stochastic process probability state change of a cloud service system, the service time distribution can be calculated, then analyze the service request, service request expected completion time, service request refused probability, and processor load performance index to measure the cloud service system performance. In the experimental analysis stage, simulation of cloud service request service is processed by the simulation tool Sim Events using discrete event simulation. According comparing average completion time and other data with mathematical results obtained by analysis, we find that the mathematical analysis results cover 90% confidence intervals of the simulation results perfectly. Finally, some research has been carried out to optimize the performance of a cloud service system, in which simulated annealing algorithm is used to seek the optimal configuration data of cloud service system for balancing the cost of system configuration and service performance to achieve the purpose of optimizing the performance of a cloud service system.Based on related research, some more fine-grained modeling conditions(considering failures and repair mechanism of the processor, processor parallel processing tasks, processor load balancing) is added in this study. Through deeply analyzing the performance and performance optimization of a cloud service system close to the actual situation of a project, a reference for the performance analysis and optimization of commercial cloud services system will be provided.