This content has been marked as final. Show 3 replies
I consider your snapshot method the best in terms of fast backup and fast restore (DBPiTR) as you talk about 3TB db.
If you use RMAN for such a big databases, then think of:
1. incremental backup with change block tracing turned on
2. "recover copy" feature of RMAN
It all depends on what the needs are for restore.
If you need a very fast restore to the last backup, snapshot is the answer.
Now if someone asks you to restore the database to a point in time before the oldest available snapshot, do you recover it from tape or you just say you cann't do it?
Or if a dbs clone is needed on a different server, what technology would you use?
Implementing rman with a tape module could be an answer.
We use Netapp FlexClone which allows us to clone a database in secs. Basically using pointers instead of physically copying files.
We keep our snapshots for one month on disk and we rarely have to go back to tape. I could not recall when was the last time we had to go back to tape to restore a database.
Maybe I can use RMAN in conjunction with Flashback db. Still looking into that...