Cloud data center adaptive efficient resource allocation method based on deep reinforcement learning

Abstract

The invention relates to a cloud data center adaptive efficient resource allocation method based on deep reinforcement learning. Firstly, an actor parameterization strategy (resource allocation) is adopted, and an operation (scheduling job) is selected according to a score evaluated by critic (evaluation operation). Secondly, updating a resource allocation strategy by utilizing gradient ascending, and reducing variance of strategy gradient by utilizing a dominant function, thereby improving the training efficiency; wide simulation experiments are carried out by using real data from a Google cloud data center. Compared with two advanced cloud resource allocation methods based on the DRL and five classic cloud resource allocation methods, the method provided by the invention has higher quality of service (QoS) in the aspects of delay and job discard rate, and the energy efficiency is higher.

Description

technical field [0001] The invention relates to an adaptive and efficient resource allocation method for cloud data centers based on deep reinforcement learning. Background technique [0002] Cloud computing has rapidly developed into one of the most popular computing models. In cloud computing, resource allocation refers to the process of allocating computing, storage, and network resources to meet the needs of users and cloud service providers. With the continuous expansion and dynamic changes of cloud data centers, many problems have arisen in cloud resource allocation, such as unreasonable resource allocation and slow response to changes. These problems not only reduce service quality, but also result in higher energy consumption and maintenance overhead. Therefore, designing an adaptive and efficient cloud data center resource allocation solution has become a top priority. However, this is an extremely challenging task due to the dynamic nature of the system state an...

AI Technical Summary

Problems solved by technology

However, A2C adopts a single-threaded training method, and the computing resources are underutilized.

At the same time, strong data correlation may occur when using A2C, because in the case of only one DRL agent interacting with the environment, similar training samples will be generated, resulting in unsatisfactory training results

Images