The Extreme Counts: Modeling the Performance Uncertainty of Cloud Resources with Extreme Value Theory

Although Cloud techniques developed rapidly in the last decade, most of the applications running on Cloud are still web-based. It is the performance uncertainty of Cloud resources that hinders the further migration of other applications, such as quality critical applications. Hence, an accurate Cloud performance model is crucial for optimized resource allocation to satisfy the quality requirements of the quality critical applications. However, the existing efforts of Cloud performance modeling focus more on the mean and variance, which cannot be leveraged to guarantee meeting the deadline miss rate of quality critical applications. To tackle the issue, a new modeling method is proposed to build performance uncertainty model of Cloud resources based on Extreme Value Theory, which can generate a proper threshold to guarantee the application's Quality of Service (QoS). Based on our experimental data and studies, the threshold calculated by our proposed model can make the average miss rate become lower than the required 5% deadline miss rate and reduced by 77% compared with the traditional modeling method. The number of times that the deadline miss rate cannot be satisfied is also reduced by 84%.