Metacomputing

Peter Kacsuk, MTA SZTAKI

The main motivation for metacomputing with respect to supercomputing was driven by grand challenge problems and by the recognition that the whole can be greater than the sum of the parts. By connecting heterogeneous computing resources and particularly supercomputers we may be able to achieve superlinear speedup in case of sufficiently complex applications like the grand challenge problems.

The talk will explain how metacomputing was developed as a natural progress of supercomputing and cluster computing. Metacomputing merges the results of these two areas with the latest achievements of network computing.

Several aspects of this new field like programming models, architectural issues, scheduling and resource management will be discussed in the talk. Special attention will be paid to computational grid concepts as well as metacomputing programming environments.

Finally, some of the most important research projects of the field both from Europe and the United States will be shortly summarised and compared.