Methods and models of resource allocation in load balancing clusters

Abstract

Modern data centers are complex and effective solutions for managing enterprises and corporations, organizing data processing and storage systems, and distributing software applications among available resources. This paper considers load balancing clusters of data centers that contain a definite set of application servers, file servers, data storage systems, and an input-output system, all interconnected by a switching system and communication channels. The goal of research is to increase the efficiency of virtualized data centers using a new methodological approach, as well as rational methods and models, to distribute their load among the available software and hardware means. Interrelated mathematical methods and models, necessary to develop specialized software for the rational planning and allocation of the data center’s physical resources on a given time interval, are proposed. These methods and models are based on the system of correct mappings of the set of request parameters into the known characteristics of the data center’s physical resources. As an optimization criterion, the maximum system performance for a certain time interval is used