A fuzzy model for detecting and predicting cloud quality of service violation

Cloud computing offers cost-effective, on-demand and pay as you use IT services to service consumers based on Service Level Agreements (SLAs). The cloud services offered by cloud providers raise the need for Quality of Service (QoS) monitoring to ensure that SLAs are maintained for accountability. As a result, service performance, which includes availability, response time and throughput, is one of the core concerns of the cloud users. This paper proposes a fuzzy-based model for QoS violation detection and prediction of service response time and duration for ensuring continuous service availability. Our experimental results show small Root Mean Squared Error (RMSE) of less than 0.50, thus confirming the greater than 99% detection and prediction accuracy for QoS violation. This fuzzy model with 16 fuzzy rules of detecting and predicting QoS violation for Software as a Service (SaaS) allows the cloud administrator to make accurate decisions for appropriate remedial action. Consequently, accountability is taken care of with a win-win benefit for both the cloud users and service providers. This contributes towards a different dimension in the measurement of quality of cloud services where accountability is taken into consideration