Scheduling Distributed Algorithms On Heterogeneous Computer Networks

Marco Alfano, Alessandro Lo Genco, Giuseppe Lo Re · WIT transactions on information and communication technologies · 1970

This study deals with a distributed scheduling organization devoted to manage stand-alone and co-operating applications on a workstation network. To this end, it considers different aspects and requirements of the applications, the operating systems, and the distributed environment. In particular, it considers the scheduling activities devoted to the dynamic allocation of processes and proposes a model for differentiating local, global, long-term, medium term, and short-term scheduling. The authors provide a theoretical scheme where the above components are conveniently organized and propose a practical implementation for a network of Unix workstations using the PVM (Parallel Virtual Machine) facilities. In particular, the paper examines the problem of the performance evaluation of co-operating processes and proposes a solution based on delay detection and accusation The scheduling solution mainly consists of a migration policy and a mechanism to be performed by the distributed components of the medium-term scheduler. This paper reports the results of some optimization problems carried out by parallel implementations as a case study.

Read the paper · More papers on PaperTik