Without the Data Restore feature, you might need duplicate backups of certain tables or dbspaces, in addition to regular archives taken. There has been no possibility to restore individual tables from a DB2 archive or user archive. To be able to restore individual tables, often a backup with DBSU UNLOAD has been taken additionally. Since it is not predictable which table or dbspace of a backup will be needed for restore, in the worse case a copy of every table would have to be available. So very often a selection of important data was made and only these were separately copied, others were missing. An additional problem lies in the fact that no forward recovery can be applied on tables.
With the Data Restore feature, you get the most flexibility using either the DB2 ARCHIVE or Data Restore BACKUP for making backups. The Data Restore RESTORE lets you restore a full database or just one storage pool from a complete archive. For information about log recovery after RESTORE of a storage pool, refer to Log Recovery. The Data Restore RELOAD gives you the possibility to reload any table from a complete archive, and there is no need to make additional copies of tables just for backup. Therefore, it is recommended to spend the resources on an improved archive process of all data rather than on making duplicate backups of specific tables.
There are still occasions where you will need to unload and reload tables:
For these kinds of operations there are several alternatives from which to choose depending on which alternative with its capabilities fits your needs best.
DB2 ARCHIVE and Data Restore BACKUP offer you the following recovery choices:
As with DB2 restore, forward recovery is possible with Data Restore RESTORE, if LOGMODE=L or LOGMODE=A is used.
In case of a DBSS failure, the database might, for example, abnormally end during startup indicating the failing LUW. During startup you can specify to omit a specific LUW. This holds for:
As described above, this relieves you from making additional unloads (duplicate backups) of single tables.
With Data Restore LISTLOG you can easily find the statement that corrupted your table and omit that (and the following commands) during recovery.
There are very few disadvantages of DB2 ARCHIVE and Data Restore BACKUP, depending on what you compare them to.
But normally, unloads have been made in addition to complete backups, so that these times in fact were added.
Therefore, in addition to offline DB2 ARCHIVE or Data Restore BACKUP, you should save the database definition files as described in Recover from System Failure.
The differences between DB2 ARCHIVE and Data Restore BACKUP are the following:
DB2 ARCHIVE
For disaster recovery on a different system, it is good to have an archive available which was taken offline, or online in LOGMODE=L, because these are consistent. Otherwise you need to be able to apply forward log recovery to get consistent. The same holds when you want to use an older archive than the latest, as described in Recover from System Failure.
You can translate every DB2 archive immediately after having taken it and keep the work files.
A Data Restore RELOAD can also be made without translation, needing more time; the translation is made internally, reading the tape twice. But this takes less time than TRANSLATE and RELOAD from translated archive.
For Data Restore RESTORE a TRANSLATE would be required; you can omit translation when restoring the database using the standard DB2 restore procedure.
Data Restore BACKUP
All of these operations can be performed directly from a Data Restore archive.
Considering the restrictions about the size of the output disk related to the size of the database, this applies only for small and medium databases (see Data Restore BACKUP).
Given the fact that we measured around double the time for an archive to disk than to tape, this is not a big advantage.
For example, you might want to keep one tape copy for table recovery, and send another tape copy to a safe place for disaster recovery.
So the most prominent difference is that a DB2 archive can be taken online, but might be translated, which can take twice as long as the archive itself. For disaster recovery in LOGMODE=A, offline archives are safer, as described in Recover from System Failure and page ***.