Modeling Distributed Computing System Reliability with DRBD

Abstract

Nowadays the great part of devices or systems we commonly use are often driven or managed by microchips and computers: cars, music players, phones, trains, planes, .... A consolidated trend of technology is to substitute mechanical with electronic parts, analogical with digital devices or controls, and so on. In this context, features like security, availability and reliability, usually summarized under the concept of dependability, are receiving higher attention. The dependability analysis, especially for what regards critical parts as computing systems or subsystems, is becoming more strategic: specific requirements and explicit or tighter constraints have to be satisfied. Even though this fact, there is a lack of suitable tools to properly model and analyze these aspects, with particular reference to reliability. To fill this gap, we propose the dynamic reliability block diagram (DRBD) modeling tool derived from the reliability block diagram (RBD) formalism. The DRBD permits to model the dynamic reliability behavior of a system through dependence models, exploited to represent dynamics behaviors as redundancy, load sharing, multiple, probabilistic and common failure mode. In this paper, the DRBD expressiveness and other capabilities, are illustrated through the analysis of a complex distributed computing system taken as example