Markov and semi-Markov models are widely used to analyse the reliability of complex computer-based systems. Dealing with the different model features is a serious problem, which leads to computational difficulties and may affect the accuracy of the reliability analysis. We discuss the classification attributes (stiffness, largeness, sparsity and fragmentedness) that are used for computer systems reliability analysis. The provided system analysis based on this classification attributes can determine the complexities of computational problem and form recommendations for the most effective methods choose. Keywords: classification feature, stiffness, largeness, sparsity, fragmentedness.
