In order to adapt to the requirements of computing power and storage resources for scientific research applications of meteorological operations, the cloud platform of meteorological infrastructure is built by using distributed computing, distributed storage and virtualization technology. And the cloud platform resource partition is designed according to different business categories, network partition, resource service objects, and each partition needs to have cloud resources. The designed cloud platform covers the functions of the IaaS layer and PaaS layer. The supporting hardware architecture is mainly distributed architecture, which provides physical computing power, virtual host and distributed storage to support meteorological big data business and much meteorological application deployment. The operation can meet the functional and performance requirements of meteorological applications for intensive infrastructure resources.