Comparison of Contention Aware List Scheduling Heuristics for Cluster Computing

In the area of static scheduling, list scheduling is one of the most common heuristics for the temporal and spatial assignment of a Directed Acyclic Graph (DAG) to a target machine. As most heuristics, list scheduling assumes fully connected homogeneous processors and ignores contention on the inter communication links. This article extends the list scheduling heuristic for contention aware scheduling on heterogenous arbitrary machines. The extention is based on the idea of scheduling edges to links, likewise the scheduling of nodes to processors. Based on this extension, we compare eight priority schemes for the node order determination of the first phase of list scheduling. Random graphs are generated and scheduled with the different schemes to homogenous and heterogenous parallel systems from the area of cluster computing. The experiments demonstrate the appropriateness of our extended list scheduling for homogeneous and heterogenous cluster architectures.