With the rapid development of cloud computing technology, unified management of multi-cloud heterogeneous resources has become a key technology to improve resource utilization and system elasticity. Aiming at the complex problem of unified management of resource operation in multi-cloud heterogeneous environment, a multi-cloud heterogeneous resource unified management platform is built to realize the global view and unified scheduling of resources. Subsequently, a combined prediction model based on ARIMA-LSTM is proposed, which effectively solves the problem of load prediction accuracy that a single model cannot more accurately perform unified management of multi-cloud heterogeneous resources. At the resource scheduling level, a container scheduling strategy based on Kubernetes is designed, combined with an improved load scheduling algorithm (LSA) to dynamically optimize the container deployment location, and a horizontal elasticity scaling strategy based on deep reinforcement learning is implemented for autonomous decision making in unified management of multi-cloud heterogeneous resources. The results show that this paper’s method reduces up to 28.32% resource wastage and 50.75% normalized cost compared to existing resource scheduling algorithms, resulting in an average CPU utilization of 92.8389%, while there is no significant increase in the model’s response time. The results prove the effectiveness of the model in this paper in the unified management of multi-cloud heterogeneous resources, providing theoretical support and practical reference for the intelligent use of resources in complex environments.
