> Agree processor speed and gbak are of prime importance to minimise
> time offline. But there is more. The sequence of events in the file
> system for a restore affect time offline and the orderly shutdown of
> applications using the database may affect the tidiness of the
> database, even with careful use of transaction management. Automatic
> schedulers have no clear way to effect an orderly shutdown. Which
> ones do? With VPN's, remote connections and telecommuting I assume
> there is no day or time when one can assure there is no application
> connected to the database. Only the database will know how many
> cooperative users are connected to it.

correct - you can automate regular backups and test restores without
interruption to users (see service I sent before)
you cannot effect a backup/restore cycle automatically IMO
A DBA can also tell if someone(s) is connected

>
> >
> > > So I thought a strategy to minimise the offline time would be to:
> > > 1. make a backup file (FileBK) of the datafile FileBD.
> > > 2. rename the datafile FileDB->FileOldDB_date
> >
> > as long as noone is updating while you backup - otherwise, rename
> >before you backup
>
> If a restore is required from an aged backup, users will have to
> know that data was lost from the start-of-backup until the-end-of-
> restore. Updating missed during backup may also be lost, just
> like all the updating in between.

an aged backup is one which was taken "last night" - with lost data
a rational restore from a backup/restore cycle should be carried out with
absolutely no loss of data.

>
> With a non-destructive or planned backup and restore the backup
> starts only after all operators have orderly disconnected from the
> database except the one requesting the backup and restore.
>

correct - hopefully SYSDBA or the OWNER of the database.

Alan