Parallel Archive Approach (ON-Archive) 

ON-Archive allows you to group one or more dbspaces in the Informix database into dbspace sets, which can be archived to different volume sets independently of each other, that is, in parallel at the same time. The advantage of performing parallel database archives is that they are more rapid than sequential archives. Therefore, parallel archives are useful for very large databases.

You must always do a logical-log file backup after a parallel archive to make sure that you can restore the database in the event of a failure. Unlike a sequential archive, a parallel archive cannot be used for a restore without the accompanying logical-log files, so it is essential you back up the log files.

If possible, you should archive sequentially. Assign several physical tape drives to one logical device if you have so much data that it exceeds one tape volume. Then you can perform the archive overnight in unattended mode. The archive is written in sequence to a number of tapes mounted on a series of tape drives.

ON-Archive does not provide a tool to check whether the total dbspace sets of a parallel archive correspond to the data volume of the entire database. For example, ON-Archive does not check whether you have forgotten to save a specific dbspace. If the database needed to be restored, you could not restore any missing dbspace. In general, you should take extra care with parallel archives.

The SAP tool for scheduling archives from within the R/3 System, Computing Center Management System (CCMS), automatically groups the dbspaces of your database into dbspace sets to enable parallel archives. Refer to DBA Planning Calendar. For more information about setting up your dbspace sets independently – not recommended by SAP unless necessary – see the Informix documentation.

Although it is possible to do a partial archive of your database (that is, only archive certain dbspaces, but not all), SAP strongly advises against this. The R/3 System is so integrated that it makes little sense to archive only one section of the database.