TITLE: Scheduling DAGs on Grids with Copying and Migration

Authors: 
Israel Hernandez, School of Informatics, University of Edinburgh, 
Murray Cole,      School of Informatics, University of Edinburgh, 


Abstract.
Effective task scheduling mechanisms for Grids
must address the dynamic and heterogeneous nature of the system by
allowing rescheduling and migration of tasks in reaction to
significant variations in resource characteristics. Migration may
be invoked when the cost of the migration itself is outweighed by
the global time saved due to execution at the new site. We extend
our previous results in this area by considering the
maintenance of a collection of reusable copies
of the results of completed tasks. We show how to reuse such copies
to improve overall application makespan. We present the Grid Task
Positioning with Copying facilities GTP/c system. We compare
the performance of GTP/c with our previous model GTP and
DLS/sr, an adaptive version of the dynamic level scheduling
static method.