Agenda WHAT ? WHO ? WHEN ? WHY ? HOW ?

Ensuring the availability of enterprise IT systems is a challenging task. The factors that can bring systems down are numerous, and their impact on various system architectures is difficult to predict. At the same time, maintaining high availability is crucial in many applications, ranging from control systems in the electric power grid, over electronic trading systems on the stock market to specialized command and control systems for military and civilian purposes. The present paper describes a Bayesian decision support model, designed to help enterprise IT systems decision makers evaluate the consequences of their decisions by analyzing various scenarios. The model is based on expert elicitation from 50 academic experts on IT systems availability, obtained through an electronic survey. The Bayesian model uses a leaky Noisy-OR method to weigh together the expert opinions on 16 factors affecting systems availability. Using this model, the effect of changes to a system can be estimated beforehand, providing decision support for improvement of enterprise IT systems availability. Keywords-Systems availability, High availability, Downtime, Bayesian networks, Noisy-OR, Expert elicitation

[1]  Max Henrion,et al.  Some Practical Issues in Constructing Belief Networks , 1987, UAI.

[2]  Ross D. Shachter.  Probabilistic Inference and Influence Diagrams , 1988, Oper. Res..

[3]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems - networks of plausible inference , 1991, Morgan Kaufmann series in representation and reasoning.

[4]  R. Cooke.  Experts in Uncertainty: Opinion and Subjective Probability in Science , 1991 .

[5]  Thomas W. Mangione,et al.  Mail Surveys: Improving the Quality , 1995 .

[6]  Nir Friedman,et al.  Learning Bayesian Networks with Local Structure , 1996, UAI.

[7]  J. Calder.  Survey research methods , 1998, Medical education.

[8]  Michal Linial,et al.  Using Bayesian Networks to Analyze Expression Data , 2000, J. Comput. Biol..

[9]  Hoang Pham,et al.  An analysis of factors affecting software reliability , 2000, J. Syst. Softw..

[10]  Marek J. Druzdzel,et al.  Learning Bayesian network parameters from small data sets: application of Noisy-OR gates , 2001, Int. J. Approx. Reason..

[11]  S. Renooij.  Qualitative approaches to quantifying probabilistic networks , 2001 .

[12]  Finn V. Jensen,et al.  Bayesian Networks and Decision Graphs , 2001, Statistics for Engineering and Information Science.

[13]  Lei Zhou,et al.  The Economic Cost of Publicly Announced Information Security Breaches: Empirical Evidence from the Stock Market , 2003, J. Comput. Secur..

[14]  Kevin B. Korb,et al.  Parameterising Bayesian Networks , 2004, Australian Conference on Artificial Intelligence.

[15]  A. O'Hagan,et al.  Statistical Methods for Eliciting Probability Distributions , 2005 .

[16]  Fredrik Johansson,et al.  Implementation and integration of a Bayesian Network for prediction of tactical intention into a ground target simulator , 2006, 2006 9th International Conference on Information Fusion.

[17]  Richard E. Neapolitan,et al.  Learning Bayesian networks , 2007, KDD '07.

[18]  Miroslaw Malek,et al.  Modeling Business Process Availability , 2008, 2008 IEEE Congress on Services - Part I.

[19]  Miroslaw Malek,et al.  Analytical Availability Assessment of IT Services , 2008, ISAS.

[20]  Robert Lagerström,et al.  Software Change Project Cost Estimation - A Bayesian Network and a Method for Expert Elicitation , 2009, CSMR 2009.

[21]  R. Jacobs.  What to research , 2009 .