An Approach to Modelling Information System Availability by Using Bayesian Belief Network

In today's era of the ubiquitous use of information technology (IT), it is expected that the information systems provide services to end-users on continuous basis, regardless of time and location. This is especially true in organizations where information systems (IS) support real-time critical operations, particularly, in the industries in which these systems must continuously operate 24x7x365. This paper presents a modified Bayesian Belief Network model for predicting IS availability. Based on a thorough review of all IS availability dimensions, we proposed a modified set of determinants. The model is parametrized using probability elicitation process with the participation of experts from the BiH financial sector. The results showed that most influential determinants of the IS availability are a timely and precise definition of the availability requirements, quality of IT operations, management and network.