Pii: S0166-5316(00)00067-5

Welcome to this special issue of Performance Evaluation, presenting selected papers of the recent 11th International Conference on Modelling Tools and Techniques for Computer Performance Evaluation (TOOLS 2000) and the Fourth IEEE International Performance and Dependability Symposium (IPDS 2000), which were both hosted by Motorola in their Galvin Center in Schaumburg, a suburb of Chicago, 27–30 March 2000. For TOOLS 2000, the aim was to further develop the theory and technology for tool-based performance and dependability evaluation of computer and communication systems. At IPDS 2000, academic and industrial researchers and practitioners met to address applied and theoretical aspects of computer performance and dependability analysis. TOOLS 2000 enjoyed 49 regular submissions, of which 21 were selected as regular papers. IPDS 2000 received 37 regular submission, of which 17 where selected for the conference. In addition to these regular papers, two invited papers and 14 short papers addressing new tools for performance and/or dependability evaluation were presented. The proceedings of TOOLS 2000, including the 14 short tool descriptions, have been published as Lecture Notes in Computer Science, Vol. 1786 (B.R. Haverkort, H.C. Bohnenkamp, C.U. Smith (Eds.)) with Springer, whereas the IPDS 2000 proceedings have been published by the IEEE Computer Society Press. In the current special issue, a selection of five papers from TOOLS 2000 and six papers from IPDS 2000 are presented. These papers have been selected by the respective program committees, and have been extended and considerably improved by their authors. An extra review procedure was invoked in the summer of 2000, and in the fall of 2000 the final papers were delivered. In the first paper selected from TOOLS 2000, Decomposition of General Tandem Queueing Networks with MMPP Input, Armin Heindl (TU Berlin) presents a novel approach to decompose tandem queuing networks with generally distributed service times, subject to Markov modulated Poisson traffic; this type of queuing network model especially finds its application in the area of high-speed networking. The inter-queue traffic is described by means of semi-Markov processes and Markov modulated Poisson processes, instead of by renewal processes (as is the case in most decomposition approaches). Thus, the correlations in the traffic streams, which considerably impact the queue performance, can be accounted for. Comparisons with simulations show the accuracy of the new approach. The evaluation of high-speed networks is also the application area aimed at by José Incera (ENST Bretagne), Raymond Marie, David Ros and Gerardo Rubino (all at IRISA in Rennes) in their paper FluidSim: A Tool to Simulate Fluid Models of High-speed Networks. When evaluating high-speed networks, it is computationally often very unattractive to simulate every packet individually. Instead, an approach can