Analysis of dependability aspects in data center infrastructures considering temperature variation and different redundancy mechanisms

Abstract

Data centers are constantly growing, driving demands for new technologies such as cloud computing, e-commerce, which forced the availability of these systems 24 hours a day, 7 days a week under penalty of great losses. On the other hand, this growth has caused a greater consumption of energy with a consequent increase in the heat dissipation by its components, which may lead to an increase in the operating temperature of these systems. In this work, we focus on the dependability aspects of a data center, particularly considering the variation in the availability of the communication infrastructure due to the variation in its temperature that is provided through the cooling infrastructure. With this infrastructure, architectures have been proposed that use different redundancy mechanisms in order to analyze how temperature variation occurs when each of these architectures is used. The dependability models were created using the Reliability Block Diagram modeling mechanisms and stochastic Petri nets. In addition, a study was prepared for the application of these models considering different scenarios. Several analyzes were carried out, such as checking the availability of the communication infrastructure as a function of temperature, considering the variation in the failure time of its components, the variation in the temperature of the communication infrastructure according to the number of active components of the cooling infrastructure and the probabilities failure of components in the cooling infrastructure considering the different redundancy mechanisms adopted.