基于图神经网络的虚拟机迁移路径优化算法  

Optimization Algorithm for Virtual Machine Migration Path Based on Graph Neural Networks

作　　者：党杜均 

机构地区：[1]仁寿智仁智慧科技有限公司,四川 仁寿

出　　处：《计算机科学与应用》2024年第9期66-78,共13页Computer Science and Application

摘　　要：随着云计算技术的广泛应用,虚拟机迁移已成为提高数据中心资源利用率和服务质量的重要手段。传统的虚拟机迁移路径优化算法在应对复杂网络拓扑和多变负载条件时,往往面临效率低下和资源浪费等问题。为解决虚拟机迁移路径优化的挑战,本文提出一种基于图神经网络(Graph Neural Network, GNN)的虚拟机迁移路径优化算法。该算法将数据中心的网络拓扑结构抽象为图模型,利用图神经网络强大的特征提取和表示能力,对虚拟机迁移路径进行全局优化。算法先构建一个包含网络节点和连接关系的图结构,将虚拟机迁移问题转化为图上的路径优化问题。其次,算法采用图卷积网络(Graph Convolutional Network, GCN)对图中的节点特征进行学习和提取,通过深度神经网络对迁移路径进行预测和优化。实验结果表明,相较于传统的虚拟机迁移算法,基于图神经网络的迁移路径优化算法在处理复杂网络环境和动态负载变化时,具有更高的效率和资源利用率。With the widespread application of cloud computing technology, virtual machine (VM) migration has become an important means to improve resource utilization and service quality in data centers. However, traditional VM migration path optimization algorithms often face inefficiencies and resource wastage when dealing with complex network topologies and variable load conditions. To address the challenges of VM migration path optimization, this paper proposes a VM migration path optimization algorithm based on Graph Neural Network (GNN). The algorithm abstracts the network topology of data centers into a graph model, utilizing the powerful feature extraction and representation capabilities of GNN to optimize VM migration paths globally. Specifically, the algorithm first constructs a graph structure comprising network nodes and their connections, transforming the VM migration problem into a path optimization problem on the graph. Next, it emp

关 键 词：虚拟机迁移 路径优化 图神经网络 数据中心 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]