Scheduling on the Grid via multi-state resource availability prediction
GRID '08 Proceedings of the 2008 9th IEEE/ACM International Conference on Grid Computing
| Hi-index | 0.00 |
Each computer system in operation today is certain toexperience faults in its operational lifetime. There ismore than one source, which can cause faults, forexample environmental factors, communication,interaction with users, or hardware and softwaredesign flaws randomly build into the system. Areliable system is the one that would continueoperation albeit in a degraded mode despite thepresence of faults. In this paper, we introduce anefficient method to schedule a tandem of real-timetasks in cluster computing with possible processorfailures. The method is based on developing anobjective function that combines different schedulinggoals. This objective function is used to guide thesearch algorithm to find an efficient solution. Theproposed algorithm consists of three terms,reliability, deadline and tasks grouping to minimizeremote communication among tasks. The proposedmethod creates a scheduling table based on theobjective function then searches this table for afeasible solution. The method is efficient because itreduces communication between different tasks bygrouping them together. This grouping processreduces application de-fragmentation as well.