Paul Randal (Principal Lead Program Manager, SQL Storage Engine) over at SQL Server Storage Engine blog, has posted today on why everybody should turn AUTO_SHRINK off for their production dbs. He enumerates three reasons why AUTO_SHRINK should be turned off. He's an authority on the SQL Server Storage Engine so I would heed what he says.
I have however one more reason to add: in case of catastrophic data loss (either through db corruption, DELETE without WHERE, DROP TABLE or TRUNCATE TABLE) from which you can't recover by restoring a backup (either because you don't have it or you don't have up to date transaction log backups or whatever), you really really *really* don't want SQL Server going in and shrinking the database files before you had a chance to recover the data. What you want to do instead is:
1. Put database in read-only mode immediately so that it's left in the state as close as possible to the state it was in in the moment of the data loss (if you experienced a hardware failure or something such your db files are now detached - just leave them like that for now)
2. Download ApexSQL Log and install it. If you need to analyze a database that's online install ApexSQL Log's server-side components on the server. It doesn't matter if you have or don't have transaction logs for the database - the application will try to recover data from what you have (besides, transaction logs can help only with the recovery of delete data)
3. Run ApexSQL Log's Recovery Wizard and chose the recovery option most adequate for the scenario. Recovery Wizard will recover all the data (including BLOB) it can still find in the database and create a recovery script.
4. Run recovery script on another database to check the data. If everything is fine - great! If there's a problem or you think that the software should have recovered more data, please contact us at firstname.lastname@example.org and we will help you out.