Storage to SAN Migration...

Main Page | Report this Page

Computers Forum Index » Computer - Databases - Informix » Storage to SAN Migration...

Page 1 of 1

Storage to SAN Migration...

Author

Message

sdhaya...

Posted: Tue Oct 13, 2009 6:10 am

Guest

Hi,

We are working with our vendor on migrating our Databases from Local
Storage to SAN. Our Database is using the Raw Partion. Vendor has
advised that they will create a similar character device link in the
system from SAN ( we have around 250 character device links) and then
copy data from the existing character devices to the new SAN devices.
Once data is copied they will rename the SAN links to the original
character device links.

My question ..This is possible... ( copying data from character
devices links, renaming character devices .. )

People who have done this type of migration can share their thoughts.

Thanks in advance.

Dhaya

Andrew Clarke...

Posted: Tue Oct 13, 2009 10:19 am

Guest

Quote:

Firstly the engine has to be offline - and backed up of course.

If the names of the devices are identical then the engine is oblivious to the
move.

Wise admins use symbolic links for the "device" names so that the engine is
insulated from movements of chunks. Sometimes when a disk crashes, the
replacement or it's partitions have a different name, so it's not unexpected
to move chunks around.

There are a few tricks you can use

1/ keeping the engine running, add mirror chunks on the new device for all
your chunks. Bring the mirrors online - some time will elapse before they are
synced. Once all the mirrors are online you can remove the old chunks.
Practice this first, there's one or two points to take care of during the
process.

2/ the restore operation can be told to rename the chunks as they are loaded.
So you can take an archive, shut down the engine, restore it into the new
location, and if it goes well you can use it running on the new chunks,
otherwise go back to the old chunks and try again. Save a copy of the
$ONCONFIG in case you need to revert to the old config.

Art Kagel...

Posted: Tue Oct 13, 2009 2:42 pm

Guest

Yes, as long as the server is offline or blocked (onmode -c block) during
the copy it will work out fine. However, your chunks should not be pointing
to the actual device links, you should be using links to those devices with
paths in another location. If you are using IDS 11.xx you can also rename
your chunk paths to links pointing to the new devices rather than rename the
devices.

Art

Art S. Kagel
Oninit (www.oninit.com)
IIUG Board of Directors (art at (no spam) iiug.org)

Disclaimer: Please keep in mind that my own opinions are my own opinions and
do not reflect on my employer, Oninit, the IIUG, nor any other organization
with which I am associated either explicitly or implicitly. Neither do
those opinions reflect those of other individuals affiliated with any entity
with which I am affiliated nor those of the entities themselves.

On Tue, Oct 13, 2009 at 2:10 AM, sdhaya <dhaya.snidhi at (no spam) gmail.com> wrote:

Quote:

Andrew Clarke...

Posted: Wed Oct 14, 2009 2:28 am

Guest

Quote:

Do check with IBM whether IBM supports raw partition on SAN system.
Several years ago, I was told by IBM that they did not support raw
partition
on SAN.

change of type - character device (raw), block device (buffered) or cooked
(plain file) is determined when the engine starts up, and any performance or
behavioural conditions are set from that boot.

So once again, it's oblivious to change of device type too. As I said,
symbolic links are the best of the lot. I'd be using the mirror or restore
tricks I mentioned, but moving the data to spaces that are referred to by a
consistently named symlink set. We use something like

/IFXCHUNKS/$engine/chunkname -> whereever

so your future admin work can be merely moving lumps around without touching
the engine config at all.

Page 1 of 1

All times are GMT The time now is Thu Dec 03, 2009 1:25 am Computers Forum Index