I published a list of the various AWR scripts a little while ago – this includes options to export and import repositories.

I can see that there would be probably be some difficulty with two repositories having the same database id, but possibly you could play around with those scripts to see if they could get you what you want.

Thanks
Suman

Thanks for that observation – you’re absolutely right.

I should have made it clear that the idea is to run the “start/end” snapshots in addition to the typical hourly snapshots.

With your permission, I’ll add a little warning for those that will use only one AWR report that covers a long overnight batch processing.

In the ‘SQL ordered by’ sections, AWR shows only the statements that are present in the ending snapshot. And a snapshot gets SQL statements from the shared pool (v$sql).

So if we have a 3 hours batch process where an expensive running query was executed only during the first hour, then there is a chance that it has left the shared pool before the end. In that case, we will not see that expensive query within our 3 hours report.

Fortunately, the ‘Time Model Statistics’ section shows the ‘sql execute elapsed time’ that covers the whole duration. And the ‘SQL ordered by Elapsed Time’ section shows the DB time for each statement.

Then it is always a good idea to check that the sum of SQL ‘DB time’ are covering a relevant part of the ‘sql execute elapsed time’.
If not, then we know that we see only a partial picture, and that an intermediate snapshot should have been useful.

Regards,
Franck.

Temporary brain glitch – there are other bits that appeared in 11.2 but you’re right that this one has been in since (at least) 10.2. I’ve corrected the note.