This research activity deals with performance evaluation and
optimization of computer systems with the help of analytical
models. Today, evaluation is an essential part of the planning,
development, tuning and comparison of computer systems as well
as production systems. Depending on the goal, the hardware
structure, the system software or user programs are the central
point. Careful modelling is therefore the prerequisite for
dependable judgements. Queueing networks or stochastic Petri
networks are often used in the modelling of hardware structures
and operating system strategies. Precedent graph models are
more suitable for modelling (parallel) user programs. The
respective models are the basis for analytical investigations.
Using these, performance values like response time,
utilization, throughput rate or speed-up can be calculated.
Researcher exchange program with
the university of Budapest, Hungary,
the Lajos Kossuth university of Debrecen, Hungary,
the university of Trier, Germany
and the university of Hamburg, Germany. Contact:G. Bolch