Multi-target cloud workflow scheduling method based on reinforcement learning strategy

Abstract

The invention discloses a multi-target cloud workflow scheduling method based on a reinforcement learning strategy. A reinforcement learning Agent is improved by using a pointer network to form an improved deep reinforcement learning algorithm so as to construct a workflow scheduling model based on a reinforcement learning strategy, so that the workflow scheduling model can be suitable for cloud workflow scheduling problems of different sizes and different types, and the generalization ability of the model is improved while relatively high timeliness is ensured.

Description

technical field [0001] The invention belongs to the technical field of cloud computing, and in particular relates to a multi-objective cloud workflow scheduling method based on a reinforcement learning strategy. Background technique [0002] In recent years, more and more scientists use workflows to build their complex applications and deploy them on cloud platforms for execution. Cloud computing is the latest paradigm of distributed system computing. Its pay-per-use and elastic resource model provides an easily accessible, flexible, and scalable infrastructure and deployment environment for fast, distributed, and efficient execution of large-scale scientific workflows. But it also brings many challenges to workflow scheduling in cloud environment. On the one hand, its elastic resource mode greatly increases the scheduling solution space. On the other hand, cloud pay-per-use makes workflow scheduling need to consider workflow execution time and cost at the same time, which...

AI Technical Summary

Problems solved by technology

However, when faced with large-scale task requests, the Q-value matrix dimension explosion problem inherent in the Q-learning algorithm requires a large amount of data storage, resulting in high algorithm storage complexity; the DQN-based algorithm uses value function approximation to solve The high-dimensional data storage problem of Q-learning has been solved, but since the reinforcement learning model is trained with a fixed-dimensional environment state vector and a single type of workflow, its model generalization ability has great limitations, and it is difficult to adapt to different sizes, Different types of workflow scheduling needs

Images