WANG Qing-jiang, XU Jian-liang. Optimization of Node Selections for Grid Jobs under De-Centralized Scheduling Frameworks[J]. Acta Electronica Sinica, 2006, 34(8): 1420-1423.
WANG Qing-jiang, XU Jian-liang. Optimization of Node Selections for Grid Jobs under De-Centralized Scheduling Frameworks[J]. Acta Electronica Sinica, 2006, 34(8): 1420-1423.DOI:
To optimize node selections for grid jobs under de-centralized scheduling frameworks
the algorithm of random multi-start hill climbing is proposed.To make multiple start-points distributed uniformly in grids
each start-point is found according to exponential increase of iteration times of selecting neighbors randomly.To reflect reasonable scheduling requirements from users
the average bounded slowdown weighted by parallel computation capacity is used to evaluate node selections.By flexibly adjusting grid workload
the algorithm of random multi-start hill climbing is evaluated comprehensively.With grid load being not too light
the algorithm can effectively optimize node selections in the whole grid.