In last ten years, building information modeling (BIM) has been rapid development. However, there are still few technologies and applications to support BIM data storage and sharing during building lifecycle. Therefore, a clouding-computing-based BIM data integration and service technology is proposed by this research. It adopts Hbase (a kind of NoSql database) to store massive BIM data on distributed nodes, which are integrated virtually by a clouding computing platform, so that all participants could get their required BIM data with a uniform interface. As one of its most import modules, a semi-structured BIM database is described in detail in this paper. By combing with attributes of Hbase and IFC model, the database addresses many problems, such as space waste due to sparse of IFC model, and lots of join calculations during information query. It's also more efficient and extensible than existed BIM databases. As a conclusion, this database sets a good foundation for clouding-computing-based BIM data integration and service technology, and is valuable for BIM application.