Uma contribuição ao Int-FLBCC explorando Consenso Fuzzy via funções de penalidade.

Abstract

The dynamic consolidation of resources in the infrastructures provided by Computing Clouds is a widely used strategy to improve energy efficiency in Cloud Computing. Determining when it is best to relocate Virtual Machines (VMs) from overloaded hosts, or that are under a very low load, is an aspect that directly influences resource utilization and the quality of service offered by the CN infrastructure. As an important step in the study and research efforts, a Systematic Literature Review was carried out through the formulation of a search string, an inclusion criterion and four exclusion criteria. At the end of this stage, eight works were selected, presenting their main characteristics and the strategies used to manage resources in the NC. In this scenario, this work aims to design an approach for the consolidation of resources in a NC environment, which considers the treatment of information related to Computing Power, Communication Cost and RAM Consumption. Based on these features, the proposed approach extends the Int-FLBCC, which consists of an Interval Valued Fuzzy Logic approach, adding a degree of reliability to the results obtained with an evaluation through consensus measures. Evaluations were carried out exploring two fronts: (i) where consensus measures and penalties in fuzzy values are explored, based on membership functions; and (ii) which considers the imprecision inherent to the Interval Valued Fuzzy Sets related to input and output variables. The resulted evaluations point to promising results in the treatment of imprecision through the functions developed in Int-FLBCC, and also, due to the exploitation of fuzzy consensus, greater reliability is generated in the treatment of the uncertainty present in the information captured from an infrastructure typical of Cloud Computing.