Filesystem backups
- From: Janine Sisk <janine@xxxxxxxxxx>
- To: oracle-l L <oracle-l@xxxxxxxxxxxxx>
- Date: Wed, 2 Jun 2010 23:59:48 -0700
I currently have my 11g (R2, EE, Linux) instance set up with datafiles on one
volume and the flash recovery area on another. One form of backup I am taking
consists of filesystem snapshots (this is on Amazon's EC2). I put the database
into backup mode, take a snapshot of the datafiles, take it out of backup mode.
Once I get my RMAN setup sorted out then it will run next, and after that I
take a snapshot of the flash recovery area.
It occurs to me that having the two snapshots separated by an hour probably
means that if I had to recover from them, there's a good chance that they'd be
out of sync; the archive logs, flashback database, etc could be newer than
what Oracle expects.
In the event that I lost my current volumes and had to run from the above
snapshots - I know that Oracle might reject some of my archive log and
flashback data, which would affect my ability to do point in time recovery.
But I think I would be ok otherwise, that is, Oracle would run and would not
notice that the files were out of sync unless I tried to do recovery.
Does this sound right? Yes, I can and will test it, but there's no way for me
to be sure that I've hit all the possible combinations of things being out of
sync.
I split this into two volumes thinking it would be a good idea to spread the
I/O around a bit, especially while RMAN is running (EBS volumes are not exactly
speedy, particularly for writes). I could consolidate it back into one volume
if this is going to be a huge problem; I can't think of any other way to
ensure that the backups are in sync.
Thanks in advance!!!
janine
--
//www.freelists.org/webpage/oracle-l
Other related posts:
