Parallel JSO-Based Approach to Task Scheduling Problems in Cloud Systems

Task scheduling in cloud systems is essential for both the customer and the service provider. One of the efficient solution methods for task scheduling problems, which is an NP-hard problem type, is the metaheuristic approach. On the other hand, due to its nature, two handicaps can often arise in metaheuristic solutions based on a random search. The first is the possibility of getting stuck in local minima due to randomness, and the second is that the solution time may increase with the increase in the problem size. Multithread /multi-process approaches can provide significant advantages in overcoming these two obstacles. For this purpose, in this study, a parallel Jellyfish Search optimizer-based approach is proposed for task scheduling in cloud systems. In the proposed approach, metaheuristic algorithms working in parallel can share the best solution with each other and shorten the solution time. Both multi-thread and multi-process versions of the proposed approach are used in the study. All versions are tested separately in CloudSim for scenarios with different task sizes and the results were shared.