Mapping a Group of Jobs in the Error Recovery of a the Grid-Based Workflow within SLA Context

Abstract

The error recovery mechanism receives an important position in the system supporting Service Level Agree- ments (SLAs) for the Grid-based work ow. If one sub-job of the work ow is late, a group of directly aected sub-jobs should be re-mapped in a way that does not aect the start time of other sub-jobs in the work ow and is as inexpen- sive as possible. With the distinguished workload and re- source characteristics as well as the goal of the problem, this problem needs new method to be solved. This paper presents a mapping algorithm, which can cope with the problem. Performance measurements deliver good evalu- ation results on the quality and eciency of the method