Informatica Economică vol. 14, no. 4/2010
19
Estimation of the Reliability of Distributed Applications Marian CRISTESCU1, Laurenţiu CIOVICĂ2 1 Lucian Blaga University of Sibiu, Romania 2 Academy of Economic Studies, Bucharest, Romania
[email protected],
[email protected] In this paper the reliability is presented as an important feature for use in mission-critical distributed applications. Certain aspects of distributed systems make the requested level of reliability more difficult. An obvious benefit of distributed systems is that they serve the global business and social environment in which we live and work. Another benefit is that they can improve the quality of services, in terms of reliability, availability and performance, for the complex systems. The paper presents results of a study conducted by the students of Economic Informatics at the University "Lucian Blaga" Sibiu, over four months. The studied population was represented by several distributed applications made under the object oriented programming techniques. This study aimed to estimate the reliability of these applications using object-oriented design metrics validation techniques. Keywords: Distributed Applications, Software Reliability, Software Metrics, Fault, Tolerance
1
Introduction Communications software services and distributed applications for the next generation must be reliable, efficient, flexible and extensible. For example, applications such as real time communication between personal computers, and flight surveillance systems must be highly available and scalable to meet the required level of reliability and performance requirements. In addition, these applications must be flexible and expandable to meet their inherent complexity and quickly respond to application requirements changes. Distributed systems are the foundation of the next-generation communications systems, including electronic commerce, satellite surveillance systems, distribution and medical imaging, real time data processing systems, flights surveillance. A reliable distributed application is defined in [3] as "a system whose behavior is predictable, in spite of partial failures, asynchrony, and reconfiguration." For example, partial failures are an inherent problem in distributed systems. The average of good working time, for the distributed system components, decreases rapidly as the number of compute nodes and communication links increases. Distributed systems consist of processes
running in parallel on heterogeneous platforms, and therefore