Utilização de modelos estocásticos para o planejamento de estratégias de políticas de manutenção em infraestruturas de refrigeração e de potência em sistemas de data centers

Abstract

With the rise of Internet-based services, there is a demand for uninterrupted operation of Data center systems, which are mission-critical resources designed to support daily services such as: cloud computing, in addition to accessing social networks, to electronic commerce and information storage, forcing the availability of these systems 24 hours a day, 7 days a week, under penalty of great losses, which requires a correct dimensioning of its communication, power and cooling infrastructures. Thus, in addition to hardware redundancy techniques applied to its components, it is of great importance to adopt correct strategies for the development of maintenance policies that can ensure uninterrupted operation. In this work, the modeling mechanisms SPN and RBD are used in the construction of models for the analysis of the impact of different maintenance policies, based on different levels of SLA, on the stationary availability of cooling and power infrastructures, considering architectures with different levels of redundancy. A case study was elaborated for the analysis of the stationary availability resulting from different scenarios, which consider different maintenance strategies from the infrastructures/architectures studied through the application of the proposed models.