| Heterogeneous computing (HC) environments are becoming popular as cost-effective alternatives to traditional supercomputing by utilizing existing computing resources connected in a network in an efficient manner. System software for supporting and orchestrating such environments has not become a standard in this community. The biggest challenge of providing a reliable framework for executing jobs in a HC environment is to provide an efficient scheduling strategy. This problem of scheduling multiple tasks on multiple hosts is NP hard. This study explored an elegant strategy for optimization and its application to HC task scheduling. The interest was in finding near-optimal solutions in tractable time, to provide a benchmark for other faster scheduling techniques. |
