Leveraging Deep Recurrent Neural Networks for Adaptive Task Scheduling and Resource Allocation in Cloud Environments

Abstract

This paper presents a deep recurrent network (DRN) for green task scheduling in the cloud. The DRN is designed to optimize resource allocation by learning the dependencies between tasks and their resources. Experimental results show that the DRN can achieve significantly better resource utilization than several state-of-the-art optimization algorithms. Due to its many benefits, including flexibility, mobility, and scalability, cloud computing has recently gained popularity. However, deploying large-scale cloud applications can be challenging due to resource allocation problems. This paper proposes a logistic regression-based deep recurrent network (LRDN) that can successfully address the cloud computing issue of green job scheduling. Our LRDN can achieve near-optimal resource allocation by predicting future resource demand and adjusting the allocation accordingly. Our LRDN also outperforms a state-of-the-art deep recurrent network in several resource-intensive scenarios. As cloud computing services become increasingly popular, the need for efficient and green task scheduling algorithms becomes increasingly essential. This paper proposes a logistic regression-based deep recurrent network (LR-DRN) for green task scheduling in cloud computing. The proposed LR-DRN can learn the scheduling patterns from historical data and accurately predict future green task scheduling results. In addition, the proposed LR-DRN can optimize the resource allocation for green task scheduling by using the predicted results. Simulation results show that the proposed LR-DRN can significantly improve the green task scheduling performance in cloud computing.