Online Dependability Assessment through Runtime Monitoring and Prediction

Computer science and engineering have, in comparison to other areas, focused on abstraction and technology, chasing the ever-changing artifacts, namely, computer and communication hardware and software. At the beginning, a focus was on functionality, a decade later on performance, next the focus was widened by including cost minimization and later quality-of-service.The result is that the time for modeling, measurements and assessment is limited and once some hypotheses are verified in part, they frequently become obsolete due new technologies, environments and applications. Not surprisingly, computer science community can boast a relatively small number of useful "laws" and principles not only due to dynamicity but also due to immense diversity of applications. Nevertheless, we are an optimistic folk, always trying to solve all the world\´s problems, generalize to death (proposing, rarely useful, general models, meta-models and meta-meta-models) as in, for example, software development process where there are no limitations so one can easily promise anything and defy any laws of feasibility or reasonableness (unlike hardware engineers who are constrained by physics).