Gfix and Gbak each read the whole database, so they each take longer

as the database gets larger - otherwise they're opposites.  

Gfix validates the physical structure of the database, making sure that 

every page is either in use or listed as free and that no pages are declared 

as used for one purpose and actually used for another.  At the record level, 

it checks that back version, fragment, and blob pointers point to back

versions, fragments, and blobs.   Gfix also validates the structure of indexes.

Gbak makes a copy of the metadata and data in the database on backup

and recreates the database on restore.  It does not use indexes on data,

so it won't detect a broken index.  A gbak restore will uncover logical errors

like duplicates on primary keys, broken foreign key constraints and other

violations of logical consistency.

If I cared about my data, I'd run gbak regularly to produce backups.  I'd run

gfix -validate from time to time looking for structural errors.  I wouldn't be

too fussy about it - the odd orphaned page or back version is just lost space

and is the expected result of a crash shutdown.  People have stopped

tripping over power cords by now, and operating systems don't crash the

way they did in the bad old days, but you still might find an orphan here

or there.  Finally, I'd restore the backup occasionally to find logical 

errors.

Ann, thank you very much for this insightful information. The clear message that I take away is that it is important to use *both* gbak and gfix, either one on their own won't give the best protection.

Many thanks.........Mick.