how to ensure the time consistent in mutiple disk groups with VVR?
Updated: 23 May 2010 | 6 comments
Hi Gang,
Here I got a question re VVR. I want to replcate our production oracle db to another host with VVR, but our oracle database is running with 3 disk groups, thus I have to set up 3 RVG. My understanding is these 3 RVG will replicate data individually. How can I ensure the time consistent with such a database to be replicated to the second host?
Say, at a point of time, I want to pause the replication, create snapshot on secondary host, but I can not pause the replication at exact the same time? what about the time consistent?
Thanks in advance.
Eric
discussion Filed Under:
Comments
You really can't guarantee
You really can't guarantee time consistency across multiple RVGs. This is another reason that maintaining a 1:1 relationship between applications and VxVM disk groups is a recommended best practice.
In your case, you'd have to quiesce the database before pausing replication on the three RVGs in order to maintain time consistency.
thanks for the update. So
thanks for the update.
So really, there is no solution for multiple disk groups with VVR? I haven't seen such a database with a couple of T bytes, running within just one disk group.
I'd have to defer to a VxVM
I'd have to defer to a VxVM expert (I'm a cluster server guy), but I don't believe there's a limitation to the size of a disk group in terms of total storage. In total number of LUNs, perhaps, but not total storage size.
As suggestet you
As suggestet you could:.
-stop/put in begin backup the db on primary
-pause replication of rvgs
-start db
-take snapshot of volumes on secondary
-restart repllication
I have working env with TBs replicating 2dg one for oracle home,engine....and one for datafiles so we can just pause replication.
The limit of object of a dg depends on the size of private region.
Regards
Marco
Thanks for the update. But
Thanks for the update.
But actually that is not what exactly I am after.
I know we can quiese the database, like put the database into hotbackup mode, even we can shut down the database, followed by the pause of replication and then snapshot on secondary.
What I am looking for is if there are any solutions from veritas vvr point of view to replicate volumes from differnet disk groups, e.g. put all volumes into one rds and replicate them time consistently.
yes for a database with a couple of TBs, being in one disk group, shouldn't be a problem, and it is very unlikely to break the limit of objects in a dg.
IBC - In-Band Control
You can't. replication is done at a RVG level, and RVGs are always a subset of a disk group. No way to ensure replication.
In version 5.0 of VxVM, the limit of objects in a dg is in the tens of thousands (number of volumes). The size limits (i.e. volume size) are virtually unreachable. Even in older versions you could specify a private region size if you were concerned about number of objects.
Principal Learning Consultant with Symantec Education Course Development
Would you like to reply?
Login or Register to post your comment.