Power-tail distributions are those for which the reliability function is of the form x for large x. Although they look well behaved, they have the singular property that E(X`) = 1 for all`. Thus it is possible to have a distribution with an in nite variance, or even an in nite mean. As pathological as these distributions seem to be, they occur everywhere in nature, from the CPU time used by jobs on main-frame computers to sizes of les stored on discs, earthquakes, or even healthinsurance claims. Recently, tra c on the \electronic super highway" was revealed to be of this type, too. In this paper we rst describe these distributions in detail and show their suitability to model self-similar behavior e.g. of the tra c stated above. Then we show how these distributions can occur in computer-system environments and develop a so-called truncated analytical model that in the limit is power-tail. We study and compare the e ects on system performance of a GI/M/1 model both for the truncated and the limit case, and demonstrate the usefulness of these approaches particularly for Markov modeling with LAQT (Linear Algebraic Approach to Queueing Theory, LIPS92]) techniques.

No abstract

No abstract

No abstract