Aiming at the performance bottleneck of the CIMISS ServiceOriented Data Storage System in operation, based on the analysis of system processes, this paper carries out the corresponding optimization on the different aspects of performance bottlenecks, analyzes the differences between traditional Share Disk and latest Share Nothing structure, designs the deployment scheme of distributed databases and the technology solution of the embedded CIMISS for data storage performance issues. In order to optimize the performance bottleneck of the scheduling process, the corresponding scheduling records are optimized. The data analysis shows that the service performance and operational stability of SOD are improved significantly; the performance of some interfaces is improved by more than 10 times; and the time efficiency of unstructured data storage is higher than 98%, when the above optimization measures are implemented. It provides stable and reliable data support for all types of meteorological service applications.