RDF/IMP and IMPX System Management Manual (RDF 1.3+)

Online Database Synchronization
Compaq NonStop™ RDF/IMP and IMPX System Management Manual—522204-001
7-16
Partial Database Synchronization Considerations
11. Rename the original table on the backup system from its temporary name back
to its original name using the SQLCI ALTER TABLE command ($DATA.
TEMP.PART becomes $DATA.TEST.PART).
12. Use the RESTORE utility with the PARTONLY option to put the loaded primary
partition of the duplicate table into the correct location. Note that MAP NAMES is
not required because the loaded partition now has the correct name on tape and can
be restored directly.
Note that the above method preserves indexes you may have on the backup system.
There is an alternate method that may be faster in some situations than the method
described above, although this method requires that you rebuild all indexes associated
with the table on the backup system.
1. Drop all indexes associated with the table on the backup system that has a partition
in need of synchronization.
2. Use the SQLCI LOAD command with the PARTONLY to load the partition directly
from the primary system to the backup system (without having to create a duplicate
table).
3. When all loads have completed, issue the RDFCOM STOP SYNCH command.
4. Create all required indexes for the table on the backup system.
Thus, if it is faster for you to rebuild your indexes than to perform the main method
above, then this alternative method can achieve synchronization more quickly for you.
Relative Tables
This method is the same as that described for key-sequenced tables above, except that
you cannot use PARTONLY to load relative tables. For relative tables, you must load
the entire table.
1. Create an entire duplicate table on your backup system with a temporary name at a
temporary location (\BACKUP.$DATA.DUP.PART).
The alternative is to create the duplicate table on the primary system at a temporary
location (\PRIMARY.$DATA.DUP.PART).
If the table whose primary partition needs to be synchronized has indexes, do not
create indexes for the duplicate table.
2. Use the SQLCI LOAD command with the SHARED option to load only the primary
partition of the duplicate table. For faster loading, do not create indexes in step 1.
3. When all loads have completed, issue the RDFCOM STOP SYNCH command.
4. If you created the duplicate tables on the primary system, then you must use the
BACKUP utility to put the entire duplicate tables with all partitions onto tape. Note
that because you only loaded the one partition, all other partitions of this duplicate
table are empty.
If you created the duplicate table directly on the backup system, skip this step.