云计算模式下SLA中的补偿策略及风险

本文基于云计算的特性,首次提出了服务水平协议(Service Level Agreement,SLA)中的三种补偿策略(分别为基于服务失败数量、基于长服务中断数量和基于累积中断时间的补偿策略),通过在SLA中约定客户容忍阀值,针对服务失败等级对客户进行分级补偿。针对三种策略的补偿成本,本文对云服务提供商造成的损失风险进行了测量和分析,得到了风险价值(Value-at-Risk,VaR)变化的特征和规律。在平均损失相同情况下,基于累积中断时间补偿策略的VaR值大于基于服务失败数量补偿策略的VaR值,因此采取基于中断时间的补偿策略风险更大。实例数据分析表明,本文提出的补偿策略是合理可行的。本文结论对于云服务提供商的补偿策略制定具有指导意义。

Compensation Strategies and Risk in SLA under Cloud Computing Mode

In this paper, we first propose three compensation strategies (linked respectively to the number of outages, the number of failures lasting more than prescribed thresholds, and the cumulative downtime)in Service Level Agreement(SLA)according to characteristics of cloud computing. By setting tolerance threshold in SLA, we provide grading compensation to customers based on the level of service failures. In order to provide a measure of risk, we compute the Value-at-Risk(VaR)for three strategies and obtain the characteristics and variation of VaR. Under the equal average losses, we can see that the compensation strategy based on the cumulative duration demonstrates a larger VaR compared with the one based on the number of failures. Hence, the former is risker. The illustrative example indicates that our proposed compensation strategy is reasonable and feasible. The results provide constructive guide for cloud service providers to make compensation strategies.