We propose a new family of job scheduling policies for parallel computer systems that can be optimized to adapt to changes in the workload. Simulation optimization is used to reveal important properties of optimal job scheduling policies. For this optimization a new approach is suggested that combines two recent stochastic optimization methods: the nested partitions method and ordinal optimizat...

We show how to evaluate the performance of solutions to finite-horizon scheduling problems where task durations are specified by bounded uniform distributions. Our computational technique, based on computing the volumes of zones, constitutes a contribution to the computational study of scheduling under uncertainty and stochastic systems in general.

PREFACE After having received my diploma from the Technische Universität Berlin in 1996, Rolf Möhring, the supervisor of my diploma thesis, offered me a research position in his group. At that time I was employed at a Berlin software company the head of which, Gert Scheschonk, strongly encouraged me to accept the offer. I accepted and in 1997 I began to work within a research initiative funded ...

Two important characteristics encountered in many real-world scheduling problems are heterogeneous processors and a certain degree of uncertainty about the sizes of jobs. In this paper we address both, and study for the first time a scheduling problem that combines the classical unrelated machine scheduling model with stochastic processing times of jobs. Here, the processing time of job j on ma...

Abstract— We consider stochastic guarantees for networks with aggregate scheduling, in particular, Expedited Forwarding (EF). Our approach is based on the assumption that a node can be abstracted by a service curve, and the input flows are regulated individually at the network ingress. Both of these assumptions are inline with the current definition of EF [1], [2]. We derive bounds to the compl...

This paper addresses the modelling and evaluation of cell scheduling policies in ATM multiplexers using stochastic Petri nets. In particular, we compare four cell scheduling policies: a FIFO policy with and without non-preemptive priorities, a threshold priority policy as recently proposed by Lee and Sengupta, and an extension thereof. The aim of this paper is twofold. First, we want to stress ...

Abstract—A case study of the generation scheduling optimization of the multi-hydroplants on the Yuan River Basin in China is reported in this paper. Concerning the uncertainty of the inflows, the long/mid-term generation scheduling (LMTGS) problem is solved by a stochastic model in which the inflows are considered as stochastic variables. For the short-term generation scheduling (STGS) problem,...

We present a first constant performance guarantee for preemptive stochastic scheduling to minimize the sum of weighted completion times. For scheduling jobs with release dates on identical parallel machines we derive a policy with a guaranteed performance ratio of 2 which matches the currently best known result for the corresponding deterministic online problem. Our policy applies to the recent...

Given a nite time interval [0; T ], a xed number of customersN , a stochastic service system with C servers, and an initial occupancy distribution (0), we show how to calculate a schedule for the customer arrival times to minimize average blocking. We have in mind a facility user who infrequently receives a burst of N items to submit where N is a random variable. If such a \bursty" user can tol...