With all due respect for the other commentators, I think it's a little more complicated than just the data sets as a record.

One problem is that any database other than the most simple ones uses objects to select and even manipulate the data (queries and functions). So just relying upon output reports--which contain, almost by definition, a SELECTION of data, and which do not normally document the data selection criteria or the actions performed on the data---is, shall we say, very limiting. The issue then becomes what reports are used to establish the "record", what they contain, and how often they are run.

To make life in the database world even more interesting, some--but not all--working databases are constructed with the ability to "roll back" (or backtrack) so that updates to the data do not extinguish the prior data (an example is the index to land title registries). 

Finally, certain types of databases (in particular, those known as CAD and CAM system) use "libraries" of related data (essentially, collateral databases), the importance of which to the "main database" varies, in some cases being absolutely essential to the value of the data sets.

It's a complicated issue, and I don't think anyone has a definitive answer (me least of all!).

If anyone is interested, the Swiss National Archives has developed SIARD, a system to archive complete databases (http://www.bar.admin.ch/dienstleistungen/00823/00825/index.html?lang=en). 

Best regards,

Fred
-----------------------------------------------------------
Frederic J. Grevin, Vice-President
Records Management Department
New York City Economic Development Corporation * www.nycedc.com
[log in to unmask] * w. 212.312.3903 * mobile 917.510.3016 * f. 212.618.5722

List archives at http://lists.ufl.edu/archives/recmgmt-l.html
Contact [log in to unmask] for assistance
To unsubscribe from this list, click the below link. If not already present, place UNSUBSCRIBE RECMGMT-L or UNSUB RECMGMT-L in the body of the message.
mailto:[log in to unmask]