On the use of OLS regression algorithm and Pearson correlation algorithm for improving the SLA establishment process in cloud computing

As businesses and individuals migrate to the cloud, the demand for cloud services increases. Therefore, the cloud providers must provide services in accordance with the expected customer’s requirements or quality of service. QoS parameters are one of the service-level agreements (SLA) key parameters that enable building a trusted relationship between the customer and the provider. A survey on QoS in cloud computing revealed that QoS management is a challenging task in cloud applications. It involves allocating resources to applications in order to ensure services based on performance, availability, and reliability. This challenge is specifically addressed in this paper making fourfold contributions. First, we identify the relationship among QoS parameters. Second, we determine the relationship between QoS parameters and the cost of cloud service. Third, we predict the cost of cloud service. Finally, we generate an SLA. We used the Pearson correlation algorithm to identify the relationships among QoS parameters. Thereafter, we used the ordinary least squares (OLS) algorithm to predict the cost of cloud service and to identify the impact of each QoS parameter on the prediction results. Then, we used Ontology to generate an SLA. The prediction model is trained and tested on a QoS dataset. When “Performance” and “Availability” are used as independent variables, the OLS cost-based predictive model results show a positive R squared of 0.8. This implies that the two selected parameters have a significant impact on the “cost”. As a result, when “performance” and “availability” are met, the “cost” is minimized. The experiment results are used to generate a Cloud SLA that is semantically represented in accordance with the standard WS-Agreement. The establishment process relies on the use of the ontological model to generate a comprehensible SLA for the cloud. The solution is based on the adoption of the Pearson correlation algorithm and OLS algorithm which allows us to generate an SLA document that meets the needs of the customer.