1) Connect new disks to the existing database servers.
2) Add this disks to ASM. DOuble check - create new group, add into it,
check that you can see, then remove new group.
3) Add disks to the existing ASM disk group;
4) Remove old disks from existing ASM group;
5) Wait until ASM complete steps 3 and 4 (I'd better do it 1 in a time - 3
and then 4)
6) Disconnect old disks.
0 downtime (except 1 and 6 may require system reboot in some cases).
2 CSR disks.
Never ever (NEVER) use physical disk name in oracle configurations. Use
symlinks only. But anyway:
- stop cluster.
- copy both disks to the new LUN's
- reconbfigure raw devices so that they will use new lun's
- start cluster
10 minutes downtime, approximately.
So, the whole work can be done with approx 2 short downtimes, 10 minutes
each
(reserve up to 2 hours just in case of troubles and double test on staging,
make hot database backup. physically copy OCRFile and CSSFile into the
normal file before doing anything).
No need to rename OCRFile and CSSFile in the system.
CSSFile and OCRFile are both 120 MB together == 20 seconds to copy. cluster
shutdown - 2 minutes, cluster startup - 5 minutes.
System reboot not required if used iSCSI; not so easy (but may be, stil noit
required) if using HBA for FC.
----- Original Message -----
From: "Michael Daskalov" <mdaskalov@(protected)>
To: <suse-oracle@(protected)>
Sent: Wednesday, October 11, 2006 2:36 AM
Subject: [suse-oracle] SPAM: Which is the best way to migrate from one
storage array to another SAN?
Hi, all
I would like to ask for you opinion about how to migrate a 50 GB Oracle
10.2.0.1 RAC database on SLES9 SP3/x84,
which is currently on a storage with TWO luns , in which there are two
partitions (one in each LUN) which are used trough ASMLib to form two
ASM disks, which are used in turn to form two different ASM Diskgroups
(DATA and FLASH_RA).
I would like to change the storage array with minimal risk and
relatively little downtime (up to 12 hours).
What would be advisable method for this operation.
I have thought about several ways for doing that:
1) (export database, create new ASM DISKS, create new DATABASE, import)
2) Create new ASM DISKS on the new LUNS from the new storage, add them
to the current diskgroups, drop the former disks from the DISKGROUPs
3) Create new partitions, shutdown the database, clusterware,
everything. Use dd if=/dev/sdOLD of=/dev/sdNEW; remove physically
/dev/sdOLD and hope that ASMLib will recognize /dev/sdNEW in the same
way (it will make the same ASMDisk appear).
4) Create new ASM DISKS on the new LUNS from the new storage; create new
ASM DISKGROUPS from then; use RMAN to copy the datafiles, and control
files; recreate redolog files
5) Create new ASM DISKGROUP for the database, then remove the old
diskgroup and recover the database using FLASH RECOVERY AREA; then
create new DISKGROUP for the FLASH_RA and change the recovery_file_dest
to the new DISKGROUP
How to change the OCR and Voting disks (I think there only dd option
exists)?
Thanks for your opinions and advices.
Regards,
Mihail Daskalov
--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting
--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting
