云环境下的服务质量SLA违例预测模型

摘要

The efficient service composition can perform using cloud computing,and SLA (Service Lever Agreement) violation prediction avoids some failure breaches and operations to ensure that quality of service satisfies service lever agreement.SLA violation prediction model of quality of service based on continuous time Markov chain is proposed.In the model,constraint specification for QoS properties is defined based on lineartime temporal logic.Key performance indexes of current component service will update in real time by using CEP.Using them as the inputs,M/M/1 queuing model implements QoS probabilistic prediction.When it exceeds threshold defined,SLA violation warning will activate.Through the practical test in the smart grids,prediction spent less time comparing with the updating time,and has high accuracy.%利用云计算实现高效服务组合,通过SLA(Service Lever Agreement)违例预测避免某些失效操作或分支,以确保服务质量满足SLA约定.提出了一种基于连续时间马尔可夫链的服务质量(QoS) SLA违例预测模型.针对QoS属性,在预测模型中定义了基于线性时序逻辑区间的质量约束规范.利用大数据分析技术CEP(Complex Event Processor)实时更新当前运行组件服务关键性能指标,并以此为输入,使用M/M/1队列模型实现QoS概率预测,当超过定义阈值,启动SLA违例预警.通过将本模型在智能电网中进行实例测试,得到实际预测时间小于服务更新时间,预测准确率较高.