This content has been marked as final. Show 4 replies
If the issue is disc space, then go via a pipe: create the pipe on the source server with mknod, and use NFS to make the file system containing the pipe visible on the destination server. Then you can start the import reading from the pipe, and start the export writing to the pipe. That way you do the whole job (all the schemas) in one exp/imp run, without staging anything on disc, and no dependency issues.
Ok, but Im affraid of network issues. I would rather do it schema by schema. Of course my given procedure is ok in this situation? as far as dependencies are concerned?
anyway I need only 5 schemas from production (which contribute 97% space used on production)