Improved Task Scheduling Strategy Using Reinforcement Learning in Cloud Environment
Sanjeev Sharma, Neeraj Kumar Pandey · 2022
Scheduling has recently been incredibly helpful in cloud computing due to the shift in resource consumption. Cloud provides many services (based on dynamism, elasticity, and uncertainty) to end-users based on their requirements which the users can access at any time, irrespective of the location, by paying for that service. The load on the cloud environment increases as the demand for applications in cloud services rises. The degradation in services (overuse of services) or the wastage of resources (underutilization) results from improper scheduling. The resources can be distributed appropriately to the different natured tasks through scheduling. The problem of resource imbalance can be avoided, and the optimization of the main execution factors, like availability, makespan time, utilization of resources, energy consumption, reliability, response time, etc., can also be optimized. Various algorithms like Heuristic, Meta-heuristic, and hybrid are proposed to justify the scheduling mentioned above. The proposed VTO-QABC is implemented and compared on the parameter throughput with different strategies. A significant improvement is found as compared to Max-Min(84.51%), MOPSO (37.82%), HABC_LJF (19.85%), Q-Learning (7.72%), VTO-QABC_FCFS(3.89%), VTOABC_LJF (3.89%) less time than MOCS.