RDF System Management Manual for H-Series RVUs (RDF 1.8)

7 Online Database Synchronization
With RDF/IMP, IMPX, or ZLT you can synchronize entire databases or selected volumes, files,
tables or even partitions while your applications continue to run.
For information about NonStop SQL/MX databases, see Chapter 16 (page 295).
Overview
The RDF online database synchronization protocol consists of the following general steps (the
details of which are discussed later in this chapter):
• Initialize the RDF configuration with the SYNCHDBTIME option.
• Issue a START RDF, UPDATE OFF command.
• Create an empty copy of the database.
• Load the data from the actual database into the copy with shared access. Because the load
operation with shared access reads through transaction locks held by applications, the
resulting file is inconsistent with respect to transactions, but it is consistent with respect to
its physical state.
• When the loads have completed and the copy of the database is on the backup system, issue
a START UPDATE command. As the updaters start applying audit, they put the backup
database to a consistent state with regard to transactions.
The RDF online database synchronization protocol can be used to synchronize entire databases
or selected parts of databases. The operations can be complex, depending upon the database
system being used (NonStop SQL/MP, NonStop SQL/MX, or Enscribe), the file types being used
(key-sequenced and relative), and whether you need to synchronize an entire database or just
selected portions. If you need to synchronize entire databases, you should first read
“Synchronizing Entire Databases Online” and “Considerations When Synchronizing Entire
Databases” (page 157).
NOTE: Entry-sequenced files and tables, unstructured files, and NonStop SQL/MP tables that
have SYSKEYs or clustering keys cannot be synchronized with the methods described below.
For those types of files, you must synchronize them offline.
Because you must run RDF with UPDATE OFF while you load the empty copy of the database,
audit data will collect in the RDF image trails before you can eventually start the updaters.
Therefore, if your database is very large, you might want to consider synchronizing several
volumes at a time (synchronize a subset of volumes and then start updating. When the updaters
are caught up, stop RDF and reinitialize RDF using synchdbtime. Then do the same with the
next subset of volumes).
NOTE: RDF does not replicate NonStop SQL/MP catalogs. Therefore, if you are synchronizing
NonStop SQL/MP tables, you might need to create NonStop SQL/MP catalogs manually on the
backup system.
Synchronizing Entire Databases Online
To synchronize an entire RDF backup database to the primary database online:
1. If RDF is currently running, issue a STOP RDF command on the primary system.
2. Purge the RDF control subvolume on both the primary and backup systems and then issue
an INITIALIZE RDF command of the following form on the primary system:
INITIALIZE RDF, BACKUPSYSTEM \system, SYNCHDBTIME ddmmmyyyy hh:mm
Overview 155