GA-Based Job Scheduling Strategies for Fault Tolerant Grid Systems

Economic grid fault tolerance scheduling using modified Genetic Algorithm

“…al. [5] concluded that the checkpoint technique provides the best performance among the four techniques. However, they did not clearly show how the final fitness value will be calculated and whether it will be based on the total execution time, which is the summation of all time consumed by all nodes to execute the schedule, or the expected schedule completion time., which is the maximum finish time over all the nodes participating in executing the schedule, known also as Makespan.…”

“…Finally, the Checkpoint technique, which, was proved to be the best working technique in [5], is very similar to the Migration technique, since it requires to know the list of supporting nodes, and the task migrates from one failing node in the chain to the next node in the chain. The difference is that the state of the executing job is saved periodically every a predefined time span, i.e.…”

“…This requires that the scheduler knows, beforehand, the nodes that will form the support chain for each executing job by saving it as a part of the schedule information as shown in [5]. The advantage of this technique is that it can work if the current executing node is completely failed, also it will not require transferring the state of the job from node to node.…”

“…al. [5] implemented four fault tolerance techniques (retry, migration, replication and check pointing) and compared their performance. Their approach estimated the schedule execution time by adding the overhead of performing the recovery of each fault tolerance technique to the total execution time of the task.…”

“…In the context of economic grids, the most widely used fault tolerance techniques are Retry, Replication, Migration and Checkpoint [5].…”