This post is more than 5 years old
2 Intern
•
155 Posts
0
1156
May 27th, 2010 10:00
Oracle ASM disks on DMX-4 migration question
We have solrais 10 hosts and Oracle 11G with ASM disks.
We are planning to migrate ASM disks (physical EMC disks) from one host to another bigger capacity hosts.
As ASM disks are allocated as Raw disks (100GB) each, immaterial how much it is used,
Is there any way to smoothly migrate this on DMX-4 any best practice, or tips, or procedure?
No Events found!
Uwe-Ud0Mk
37 Posts
0
June 2nd, 2010 07:00
You have multiple options.
1) ASM:
Present the new HVs to all hosts and have ASM move the data for you. Then remove the old LUNs.The downside is, your are not in control and you have to depend on a DBA to do this for you. So if he or she gets it wrong, the blame game will start. Also, if you yank a single LUN from a ASM group, it will destroy the entire group. We had this before and representing the LUN back didn't help it and we had to do a full restore. IMHO, you should only do this if you can't afford the downtime (usually a lame excuse because how would you ever upgrade the database) and you have full confidence in your DBAs.The advantage to this approach is that you get a chance to change volume sizes.
2) OpenReplicator:
Sync the data between the old LUNs and the new one with OpenReplicator. OpenReplicator is not free but I have seen EMC offer it for free temporarily for data migrations. You are in control but you need downtime to pull the old and present the new HVs.I think, technically, you can change HV sizes too (go larger only, can't go smaller) but I don't know how ASM would deal with that.
3) dd
This is the poor man's approach. Present the new LUNs and dd the data over bit by bit to the new LUNs. I don't know when this would be the best alternative but it is a viable option. Source and target LUNs have to identical in size and the database has to be down while the data is copied.
dynamox
9 Legend
•
20.4K Posts
0
May 27th, 2010 13:00
if this is RAC ..can you add new nodes to the RAC and evict old nodes ?
Joe_Ab
2 Intern
•
227 Posts
0
May 31st, 2010 03:00
Why not backup/restore ?
dynamox
9 Legend
•
20.4K Posts
0
June 2nd, 2010 10:00
i don't think the intend is to migrate to another array, but to bigger hosts while retaining the same backend.so not sure about your data migration options ?
KD8EWE
2 Intern
•
155 Posts
0
June 24th, 2010 12:00
Thank You Uwe, You hit the nail with options
Dynamox yes we are moving to better storage.
emc_troy
2 Intern
•
142 Posts
0
June 24th, 2010 21:00
Like dynamox I thought you are not moving to new storage, but since you do, to add to Uwe (option 1):
The on-line procedure to migrate ASM to new storage is:
-- Add new disks:
alter diskgroup DiskGroup1 add disk 'ORCL:DB0009', 'ORCL:DB0010', 'ORCL:DB0011', 'ORCL:DB0012';
-- Rebalance:
alter diskgroup DiskGroup1 rebalance power 11;
-- Remove the old disks:
alter diskgroup DiskGroup1 drop disk DB0001, DB0002, DB0003, DB0004;
-- Rebalance:
alter diskgroup DiskGroup1 rebalance power 11;
After 10gR2 you can also do it with a single command:
ALTER DISKGROUP DISKGRP01
ADD DISK 'ORCL:VOL2'
DROP DISK VOL1
REBALANCE;