A new Grid scheduler with failure recovery and rescheduling mechanisms: Discussion and analysis

摘要

Computational Grids (CGs) have become an appealing research area. They suggest a suitable environment for developing large scale parallel applications. CGs integrate a huge mount of distributed heterogeneous resources for constituting a powerful virtual supercomputer. Scheduling is the most important issue for enhancing the performance of CGs. Various strategies have been introduced, including static and dynamic behaviors. The former maps tasks to resources at submission time, while the latter operates at run time. While static scheduling is unsuitable for the dynamic Grid environment, scheduling in CGs is still more complex than the proposed dynamic solutions. This paper introduces a decentralized Adaptive Grid Scheduler (AGS) based on a novel rescheduling mechanism. AGS has several salient properties as it is; hybrid, adaptive, decentralized, and efficient. Also, AGS is a robust mechanism as it has the ability to; (i) detect resource failures, (ii) continue its functionality in spite of the failure existence, then (iii) recover back. Moreover, it integrates both static and dynamic scheduling behaviors. An initial static scheduling map is proposed for an input Direct Acyclic Graph (DAG). However, DAG tasks may be rescheduled if the performance of the allocated resources changes in away that may affect the tasks' response time. AGS overcomes drawbacks of traditional schedulers by utilizing the mobile agent unique features to enhance the resource discovery and monitoring processes. Experimental results have shown that AGS outperforms traditional Grid schedulers as it introduces a better scheduling efficiency.