Abstract: The variety of databases structures as well as multiple data correlation on database index sets are major barriers on data sharing between heterogeneous databases. Most research work currently centralized on coherence access methods over heterogeneous databases, however, those methods can’t establish a system in good scalability. In this paper, a new resolution is presented, which firstly find out general characteristics between many heterogeneous databases, and then the complexity of each database is reduced by using dimension reduction methods, so that data\|sharing becomes more flexible with a good scalability. In addition, the principal component views（PCV） reduce data correlation between shared\|databases.