Abstract: With the rapid development of the meteorological industry， the support of various meteorological departments for information technology is increasing. The development of meteorological business needs to rely on various information systems and infrastructure. At present， the provincial meteorological data center carries a variety of meteorological services. However， due to the wide variety of services， the existence of a variety of heterogeneous devices， and the lack of an effective centralized hardware monitoring and warning platform， it has brought great difficulties to the system operation and maintenance personnel. Aiming at these problems， this paper analyzes the differences and monitoring status of heterogeneous devices in multi-source complex meteorological data environment， and designs a comprehensive hardware monitoring platform with “whole process”， “centralization” and “visualization”. The technical architecture of Mycat+MySQL+Haproxy+Keepalived is adopted to build a highly available， high-performance and scalable distributed database cluster architecture， which realizes the collection， processing， storage， user-defined warning and display of hardware monitoring data. At present， the system has been used in Gansu Meteorological Data Center， which is stable operation， safe and reliable， and can help the operation and maintenance personnel to locate the system fault in time， thus improving the efficiency of monitoring and operation and maintenance and providing efficient meteorological data services for social users.

Key words: distribution, hardware, monitoring, meteorological data, visualization