Uma Abordagem Flexível para Consolidação Dinâmica de Servidores na Computação em Nuvem Explorando Lógica Fuzzy Valorada Intervalarmente

Abstract

Considering the characteristics and scalability of typical computing cloud environments, one of its main demands is efficient resource management. Among resource management techniques, the dynamic consolidation of the virtual machines (VM) employed stands out. This consolidation has proved to be an effective technique, with a high potential to optimize the use of resources and energy efficiency of data centers employed in computing clouds. The consolidation of VMs is structured in four main steps: determination of overloaded physical machines, determination of underloaded physical machines, selection of VMs for migration, and allocation of VMs. The modeling of these phases, and consequently their impact on resource management, directly influence aspects of their use, such as energy efficiency, SLA and, consequently, QoS. Although several strategies are being proposed to perform resource management, an approach that allows the consolidation of different scenarios and/or applications of cloud computing is still unavailable. With this motivation, this work proposes using strategies based on fuzzy logic in decision making when managing resources in cloud computing, which allow a preventive treatment of the uncertainties associated with the variables of the computational environments employed. More specifically, this work has as its central objective the design of a flexible approach employing interval-valued fuzzy logic to address the issues of dynamic server consolidation in cloud computing. The approach also envisages the employment of new aggregation operators in the fuzzy system’s inference process, representing unions and intersections. The proposed approach also considers using methods for ordering intervals to support cases where conventional methods do not meet. The results achieved using the Int-FLBCC approach allowed good energy efficiency levels while preserving the cloud computing infrastructure’s SLAs.