A Logical Structure based Reliability Evaluation Model for Information Systems and its Application

In order to evaluate the reliability of information system with hybrid structure, a logical structure based reliability evaluation model for information systems is proposed for assessing hardware and software reliability for information system, in which the hybrid structure is used to architecture the reliability model, and the reliability value is used to quantitatively describe the reliability of information system. And then the NGBOSS architecture and its core processes is described. Furthermore, a reliability evaluation result is obtained by collecting and analyzing the date of two sub-processes of the core from China Mobile Group Guangdong Co., Ltd. Introduction The reliability of information systems include a number of quality characteristics, such as correctness, maturity and fault tolerance, etc. It takes the frequency of problems into account and involved running experience and the influence of errors directly. The measure of reliability refers to performed quantitative measurement to the given property which belongs to the system and can affect the system, including the reliability of hardware and software [1]. NGBOSS (Next Generation Business Operation Support System) is China Mobile’s new generation operation support system. One of the major differences between NGBOSS and former business operation support system is ABM (Account Balance Management), which established a unified view for pre-pay mode and post-pay mode from a management perspective by the new mode of billing system. The burden of huge user date, however, has created a great challenge to the billing system. The research of reliability in billing system of NGBOSS is of great significance under such a background [2]. Fault type of information system and the evaluation method of reliability The failure of information systems is divided into hardware failures and software faults in general. The reliability evaluation of information system, therefore, is divided into hardware and software evaluation. There are significant differences between the two evaluation methods as they failed for different reasons. Hardware failure is usually caused by the aging of components, and software failures are often due to design defects or overloading the computational load, etc. Therefore, it is needed to make a distinction on the type of fault before to design the evaluation method of reliability in information. Hardware failure of information system and the reliability evaluation. There are a variety of reasons that produce hardware failure including hardware unstable, aging of components and hardware affected by adverse environment. For the information system threatened by hardware failures, the reliability can be evaluated from the association between single-unit hardware and the other hardware, in the general, the hardware structure can be treated as a hybrid structure of series and parallel [3]. Fig. 1 shows three types of hardware connections in general. 2nd International Conference on Management Science and Industrial Engineering (MSIE 2013) © 2013. The authors Published by Atlantis Press 733 (a) Series structure (b) Parallel structure (c) Hybrid structure Figure 1(a) represents the hardware series structure, in which each component’s failure can cause the failure of entire structure. The reliability of this structure is the product of the reliability of each component. R = R(A) * R(B) * R(C) (1) Figure 1(b) represents the parallel structure of hardware, in which each component failure will not cause a system failure, and this structure has the characteristics of disaster tolerance. The reliability of this structure can be represented as the following formula. R = 1-(1-R(D)) * (1-R(E)) * (1-R(F)) (2) Figure 1(c) can be seen as a hybrid structrue, and the reliability of the system described in it can be represented as the following formula. R = R(A) * R(B) * [1-(1-R(C)) * (1-R(D)) * (1-R(E))] (3) Software failure of information system and the reliability evaluation. The quality of software is connection with a lot of factors, including the development environment of software system, the complexity of functionality, the level of production management, the quality of the developer and the material of software production (the development time, funding etc.). A narrow understanding of the reliability of software system is that the probability of failure-free operation of software in a specific operating environment and a specific period of time. The reliability evaluation of software can be carried around different perspective, one is evaluated by the structure of software, and the other is evaluated by a model which is based on reliability evaluation. The method used by the former is similar to the method of hardware reliability analysis [4], and the latter is though selecting the evaluation model to fit the software, and to use the definitive algorithm to obtain the reliability indicators of software. The determinants of reliability of software is the error of software which is connection with input data, is a function of data and the state of internal program, which can be measured by the probability of fault (the probability density of failure in the dimension of time), or reliability (the probability of failure does not occur within the time interval [0, 1]) [5]. Addition to the single process reliability, the more important indicator to calculate the reliability in software is the reliability of the software as a whole, which requires analysis of the structure of the software. It should be treated differently when calculate the reliability between the different information system, since different systems concerns different focus. An application of reliability evaluation to billing system of NGBOSS The architecture of billing system in NGBOSS is shown in figure 2.