Dual-mutation mechanism-driven snake optimizer for scheduling multiple budget constrained workflows in the cloud

Abstract

Scheduling workflows in cloud computing is to find the most appropriate mapping from a series of dependent tasks to a set of available virtual resources so as to minimize or maximize some user concerned objectives, which is very significant to a sustainable and high-efficient cloud data center. However, it still faces challenges since its NP-hardness and the diversified requirements of both the cloud service consumers and cloud service providers need to be satisfied at the same time, especially in addressing multiple applications requested simultaneously. This work proposes a dual-mutation mechanism-driven snake optimizer for scheduling multiple workflows in the cloud for minimizing the makespan of each workflow under user pre-defined budget constraints. Firstly, a continuous optimization algorithm, namely Snake Optimizer (SO) is adopted into discrete optimization, i.e., workflow scheduling. Secondly, a task execution order aware fitness function is designed to reduce the gaps or waiting time between parent and child tasks within a workflow and thus reduce the total execution time of the workflow. Besides, we analyze the existing snake optimizer and adjust the parameters corresponding to different evolutionary stages to adapt to our considered problem. Finally, a dual-mutation mechanism is developed by introducing a non-improvement iteration number for each snake and applying a standard bit mutation operation to prematurely converging snakes and the individuals randomly selected from the remaining snakes so that the population diversity can be enhanced and more potential solutions can be explored. Experimental results on a set of real-world scientific workflows show that our proposed algorithm is of great superiority in constraint satisfiability, meaning that compared with its peers, it is always the first to find feasible solutions.