DDPG: Cloud Data Centre Hybrid Optimisation for Virtual Machine Migration and Job Scheduling
Gurpreet Singh Panesar, Raman Chadha · 2023
The rising energy consumption in cloud data centres (DCs) has made efficient resource management crucial, since it leads to higher prices for cloud customers and negative environmental effects. Virtual machine (VM) consolidation techniques have been put out in an effort to overcome this difficulty. Their goal is to minimise energy use by reallocating VMs through migration. In this study, we propose a unique method for optimising resource utilisation and energy consumption in cloud DCs through virtual machine consolidation, based on the Deep Deterministic Policy Gradient (DDPG) algorithm. The suggested DDPG approach takes use of the DDPG algorithm's capacity to manage continuous action spaces, which makes it a good fit for handling challenging VM migration and task scheduling issues. By taking into account two goal functions in cloud DCs at the same time, DDPG seeks to find the best solution for both work scheduling and VM placement. In particular, when DDPG is used in place of the comparable approaches, there is a noteworthy 62 Kwh reduction in energy usage and an astounding 78% drop in resource utilisation.