Developing methodology of evaluation computer system reliability and performance

Aim of project
The project relates to dependable computing problems. The planned research is aimed at developing a special software environment for the analysis of critical problems in computer systems, validating its behaviour, etc. In particular we will develop efficient fault injection tools, which will allow analysing fault effects in real applications. The theoretical ideas and developed tools will be verified for various classes of applications (calculation oriented and real-time). The developed simulation environment will be enhanced with various statistical tools co-operating with system logs, etc. The developed tools will be used to optimise test scenarios, reliability and availability. New error detection and error handling techniques will be proposed and verified. To verify the proposed methodology and developed tools we will create a benchmark of various applications covering different operational and performance profiles. For these applications we will generate representative input data, checked with various coverage analyses (including program mutation techniques). Various measures to evaluate different properties of software applications will be proposed. This research will be extended with a new methodology of collecting data on reliability and performance of computer systems (data exploration and statistical correlations analysis techniques).

Expected results
The results of the project will comprise specialised software tools targeted at evaluating system dependability, analysis of the propagation of fault effects in computer systems (workstations and embedded systems), methods of improving system testability and fault tolerance. The elaborated concepts will be verified n many applications. The results of this research will be published in scientific papers as well as in special technical reports.