As described above, we removed the schema information as well as the data of each associated RDB, and stored them in another arbitrary RDB. Then, through that arbitrary RDB, we were able to manipulate those original RDBs, which is the first step to the database virtualization technology. Through this experimental process, however, we noticed some information pieces that were not consistent among the databases. They were of the data type and the constraint condition. Regarding the data type, several different kinds of data exist, each of which was unique to each RDBMS. For that reason, we were unable to associate the data type perfectly with that of other DBMS. We therefore compromised and used a similar data type for them.