Dynamic self-scheduling for parallel applications with task dependencies
Proceedings of the 6th international workshop on Middleware for grid computing
On the Origin of Grid Species: The Living Application
ICCS '09 Proceedings of the 9th International Conference on Computational Science: Part I
On the Feasibility of Dynamically Scheduling DAG Applications on Shared Heterogeneous Systems
Euro-Par '09 Proceedings of the 15th International Euro-Par Conference on Parallel Processing
The Living Application: a Self-Organizing System for Complex Grid Tasks
International Journal of High Performance Computing Applications
An efficient weighted bi-objective scheduling algorithm for heterogeneous systems
Euro-Par'09 Proceedings of the 2009 international conference on Parallel processing
Resource management of distributed virtual machines
International Journal of Ad Hoc and Ubiquitous Computing
Hi-index | 0.00 |
The execution of distributed applications on the Grid is alreadya reality. However, as both the number of applications grow andGrids increase in scale, the efficient utilization of the availablebut shared heterogeneous resources will become increasinglyessential to the Grid's successful maturity. Furthermore, it isunclear whether existing Grid management systems are capable ofmeeting this challenge. The EasyGrid middleware is a hierarchicallydistributed application management system (AMS) that is embeddedinto MPI applications to autonomously orchestrate their executionefficiently in computational Grids. The overhead of employing adistinct AMS to make each application system aware brings at leasttwo benefits. First, the adopted policies can be tailored to thespecific needs of each application, leading to improvedperformance. Second, distributing the management effort amongexecuting applications makes Grid management more scalable. Thisarticle focuses on scheduling policies of an AMS for a particularclass of application, describing a low intrusion implementation ofa hybrid scheduling strategy designed to elicit good performanceeven in dynamic environments such as Grids. Usingapplication-specific scheduling policies, near-optimal runtimeshighlight the advantages of self-scheduling when executing one ormore system aware applications on a Grid. Copyright © 2006John Wiley & Sons, Ltd.