A Cloud Computation Job Co-location Execution Method and System Based on Reinforcement Learning

Abstract

This invention proposes a method for co-located execution of cloud computing jobs based on reinforcement learning, comprising: constructing a policy model using a neural network; setting reinforcement learning parameters for the policy model; training the policy model using reinforcement learning based on individual and combined jobs of a cloud computing task; generating batch jobs for the cloud computing task; obtaining the execution strategy for the batch jobs through the policy model; and allocating and executing the batch jobs according to the execution strategy. This invention also proposes a system for co-located execution of cloud computing jobs based on reinforcement learning, and a data processing device for co-located execution of cloud computing jobs.

Images