Experience in using MEADEP

This paper reports some of the authors' experience in using MEADEP-a newly developed measurement-based dependability evaluation tool that includes both data analysis and modeling functions. Several issues are discussed: identification of time between outages and time to repair distributions; need for more graphical model forms; and consistency between parameter estimation and model evaluation algorithms. The identified distribution functions are valuable for detailed analysis and realistic modeling. The significance of discovering the inconsistency between the failure rate estimation and the availability evaluation algorithms is not limited to only MEADEP because both algorithms are commonly used. The need for more graphical model forms and the demand for better user interfaces are driving improvements for MEADEP. These issues provide insights into the roles and future directions for this kind of tools.