Constraint-Aware Resource Matching for Virtual Machine Placement in Cloud Environments

Abstract

With the continuous expansion of cloud platform scale and the increasing complexity of business workloads, resource matching and placement in virtual machine arrival flow scenarios have become a critical factor affecting the operational efficiency and service quality of cloud infrastructure. Addressing the shortcomings of existing methods, such as insufficient understanding of dynamic requests, inadequate characterization of node states, and limited ability to integrate constraints, this paper proposes a unified modeling method for placement recommendation centered on virtual machine resource allocation tasks. This method focuses on the adaptation relationship between virtual machine requests and candidate nodes, incorporating request priority information, node dynamic state information, and deployment constraint information into the resource placement decision process, constructing a recommendation framework consisting of request encoding, node encoding, feature interaction, constraint modeling, and ranking optimization. During data processing, continuous arrival flow samples are constructed based on publicly available cloud platform virtual machine trajectory data, and a supervision signal is generated by combining resource demand, node remaining capacity, and deployment feasibility relationships, enabling the model to learn effective matching patterns between requests and nodes. The results show that the proposed method can better improve the ranking quality of candidate nodes and the ability to identify target nodes, enhancing the model's ability to express dynamic resource relationships in complex cloud environments while ensuring placement feasibility. Overall, this method provides an effective solution for intelligent resource scheduling in virtual machine arrival flow scenarios, taking into account task importance, node carrying capacity, and actual constraints.