Improved Fault Tolerant Elastic Scheduling Algorithm (IFESTAL) [ READ ]
Sukhbir Singh and Raman Kumar
The existing fault tolerant models focus on the failure of host. If a host fails, the copies of tasks on this host will fail to finish. At one time instant, at most one host may encounter a failure, that is to say, if the primaries of tasks fail, the backups can always finish successfully before another host fails. Failures can be either transient or permanent, and are independent, affecting only a single host. In this paper, an attempt will be made to propose an improved algorithm to tolerate multiple host failures which will be tested using guarantee ratio, hosts active time, ratio of task over host time.
Sukhbir Singh and Raman Kumar
The existing fault tolerant models focus on the failure of host. If a host fails, the copies of tasks on this host will fail to finish. At one time instant, at most one host may encounter a failure, that is to say, if the primaries of tasks fail, the backups can always finish successfully before another host fails. Failures can be either transient or permanent, and are independent, affecting only a single host. In this paper, an attempt will be made to propose an improved algorithm to tolerate multiple host failures which will be tested using guarantee ratio, hosts active time, ratio of task over host time.
