Cuckoo search based workflow scheduling on heterogeneous cloud resources

Cloud is emerging as a utility paradigm for effectively executing large scale distributed workflows with huge computation and data transfer requirements. The heterogeneity in cloud resources makes it challenging for scheduling workflow tasks effectively. The goal is to find an optimized mapping between individual workflow tasks and cloud resources, based on various quality of services. The dependencies in workflows make it challenging for scheduling task with precedence constraints. The tasks need to be scheduled in such way that it minimizes the makespan and execution cost for workflow. This paper proposes a workflow scheduling algorithm based on cuckoo search for finding a schedule for workflow tasks. The tasks are added to execution list after prioritizing them according to the length of execution and dependencies. Cuckoo search algorithm is adapted to find an optimal mapping of workflow tasks on cloud resources. The workflow is executed within a user defined deadline while optimizing the overall makespan. The results are compared with the meta-heuristic population based Particle Swarm Optimization and heuristic technique Heterogeneous Earliest Finish Time. The results of the proposed algorithm indicate that it outperforms the approaches considered for comparison.