The estimation of probability characteristics of cloud computing systems with splitting of requests

Abstract

Growing popularity of cloud services is explained by many advantages of them. The accessibility, flexibility, scalability, ease of management, the relatively low cost of implementation can be listed among the main advantages. The demand for cloud services with the ability to change one cloud service provider to another one without any significant cost for a user result in a high competition between cloud providers. Due to this reason, it became important to find the optimal performance measures of cloud systems. These measures, on the one hand, must meet all the requirements of Service Level Agreement (SLA), on the other hand, do not lead to excessive costs for provider. The paper presents the evaluation of the main service quality characteristics of cloud systems, including formulas for variance of residence time in the synchronization buffer. For the analysis of a cloud system, fork-join queues with corresponding methods of its approximation were used.