All checkpoint files, journal files, and dump files can reside on the same device because journals and dump files are useful in recovery only if the associated checkpoint is also available. By default, the install program places journal and dump files in the same location as the checkpoint files.

If you lose the Dump location or fill the file system, Ingres will fail as it attempts to update the "dump copy" of each database configuration file. You must recreate these directories, copy each database configuration file from the database root data directory and restart Ingres.

If you lose the Journal location, any journaled databases using it will cause the archiver to shut down. You then have between a minute and an hour (before the transaction log file fills up) to disable journaling on each affected database. Then you must replace the failed component and rebuild the directory structure. To restart journaling you must run an offline checkpoint on each database. If you fill the journal file system, free disk space and restart the archiver.

If you lose the Checkpoint location, the online system is not affected. However, you now have no local backup for the databases using this location. Replace the failed component, rebuild the directory structure, and hope that you do not need a checkpoint until the next one is created. If you fill the checkpoint file system, the checkpoint process will fail. You must free sufficient space before the next checkpoint.

The Dump location requires negligible bandwidth. The Journal location requires a moderate amount of writing bandwidth during each archiver run. The Checkpoint location requires an enormous amount of writing bandwidth as each checkpoint is run.