Fault Tolerant Scheduling of Workflows using Improved Check Pointing Technique

  • Kanagaraj K, Swamynathan S

Abstract

Workflow is a set of tasks arranged according to the data and control dependency.  Fault tolerant workflow scheduling is difficult to achieve due to its complex architecture and dynamic resource requirements. However the emergence of cloud has provided new hope for ensuring fault tolerance in workflow scheduling. The elasticity in hiring and releasing resources in the cloud helps to reduce the node failures when scheduling workflows. However workflows with task and data dependencies introduce various timing and consistency problems. To enable fault tolerance, the state of an executing program is saved on a stable storage for recovering the system even after failure. The time at which the state is stored is called as check point. Adding numerous check points will improve the reliability, but affects the performance. On the other hand the less number of check points will need more time to recover from failure. In this paper, an Improved Check Pointing Technique (ICPT) is proposed. The ICPT determines the optimal check point interval based on the mean time between failure and mean down time. It reduces the check pointing overhead by 33% and identifies a system with maximum availability for storing the check pointed image. Experimental results shows that the proposed system ensures high fault tolerance during workflow scheduling and reduces the check pointing overhead when compared to the existing techniques  

Published
2020-04-20
How to Cite
Kanagaraj K, Swamynathan S. (2020). Fault Tolerant Scheduling of Workflows using Improved Check Pointing Technique. International Journal of Advanced Science and Technology, 29(7s), 1242 - 1251. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/10697