Evaluating Cloud Service Reliability: A Novel Approach Using Almost Stochastic Dominance

This study introduces an innovative methodology for assessing cloud service reliability by applying Almost Stochastic Dominance (ASD) theory to cloud server performance data. We analyze outage information from major cloud service providers, collected via the Cloud Harmony platform and Panopta's monitoring services, spanning July 2022 to February 2023. Our approach employs both Almost First-Order Stochastic Dominance (AFSD) and Almost Second-Order Stochastic Dominance (ASSD) methodologies, utilizing a multi-threshold framework with ε values of 0.0014, 0.0405, and 0.0732. This allows for a nuanced evaluation of service performance across varying levels of risk tolerance. Key findings reveal a highly competitive market at the strictest threshold, with performance disparities emerging as thresholds are relaxed. Notably, one provider demonstrates superior risk-adjusted performance under ASSD analysis. This research contributes a more robust and flexible analytical framework to the field, offering valuable insights for both consumers selecting cloud services and providers seeking to enhance their market position. Our findings underscore the importance of considering the full distribution of outcomes in cloud service evaluation, particularly for critical applications, and highlight the utility of ASD in capturing subtle performance differences often overlooked by traditional analysis methods.