Re: Moving a 250gig database from SUN to HPUX
- From: Daniel Fink <Daniel.Fink@xxxxxxx>
- To: oracle-l@xxxxxxxxxxxxx
- Date: Tue, 25 May 2004 15:19:39 -0600
This is just a random thought, never tried it personally, just thought of it,
so really put it to the test...
It seems that the logical thing to do is to start with any/all tablespaces that
are read only. As long as you are assured that they
will not change between the export time and the go-live time. Let's assume they
do.
1) Export/import all read-only objects to new db.
2) Start of window
3) Export/import all 'small' tables (need to define small)
4) Export/import all 'medium' tables & begin index creation of previously
imported small tables (start with constraint indexes, then
do other indexes).
5) Export/Import all 'large' tables & begin index creation of previously
imported medium tables (start with constraint indexes, then
do other indexes).
6) Create constraint required indexes for large tables.
7) Create non-constraint required indexes for large tables.
Why? Here's my thinking... exp/imp is going to be read-intensive from the
source db, heavy on the network and write-intensive on the
target db. So, while you are exporting/ftping/importing, your cpu and memory
usage is probably pretty small, why not use it to do
the sorting that index creation needs? Granted, the io subsystem is going to
get hammered (like Dan after a couple of drinks of
BARC) and you will need to make sure you import just the data & grants.
Other random thoughts...
1) make your log_buffer (no more than 3m I think) and redo logs large enough so
that you don't get a lot of log associated waits.
Your concern right now is not recoverability, but throughput. Fix the settings
later for real life operation.
2) create all your indexes unrecoverable. If you have to perform any sort of
recovery, your 36 hour window is 'out the window'
anyway. After you are up and running, make sure all the appropriate objects are
recoverable.
Again, these thoughts have just come to me, so they have not been tested. Any
and all other thoughts are welcome. YMMV, LBWWCTR...
Daniel
Guerra, Abraham J wrote:
> Content-Type: text/plain;
> charset="us-ascii"
> Content-Transfer-Encoding: quoted-printable
> Hello group,
> =20
> Has any of you moved a 250+ database from SUN into HPUX? I have a
> window of 36 hours to do this. I'm planning to divide the database into
> multiple imports/exports and multi-tasks them.
> =20
> Any ideas or tools out there that would be a better approach?
> =20
> Your ideas/comments are appreciated.
> =20
> Abraham Guerra
> American Family Insurance
>
>
>
> -- Binary/unsupported file stripped by Ecartis --
> -- Type: image/jpeg
> -- File: Notebook.jpg
> -- Desc: Notebook.jpg
>
>
> ----------------------------------------------------------------
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> ----------------------------------------------------------------
> To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
> put 'unsubscribe' in the subject line.
> --
> Archives are at //www.freelists.org/archives/oracle-l/
> FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
> -----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
