Scenario
=====
Oracle 9i Enterprise Edition (9.2.0.8)
Windows Server 2003 (32 bit)
--- to ---
Oracle 11g Enterprise Edition (11.2.0.3)
Windows Server 2008 Standard R2 (64 bit)
Hi to all
I'm doing a upgrade from 9i to 11g and i am using native imp/exp to migrate those data.. For my 1st round of testing, I have done the following:
1) Full DB Export from 9i. exp system/<pwd>@db FULL=Y FILE=export.dmp log=export.log
Encountered warning "EXP-00091: Exporting questionable statistics." (On hindsight, I know that I need to set the characterset as per my server before exporting) Nevertheless, I proceeded on with this 8.4GB dmp file that has the warning "EXP-00091: Exporting questionable statistics." The characterset in 9i is US7ASCII. My export took 1 hour, my 9i is actually only a small 26GB.
2) Full import to 11g. My 11g is a newly created DB with characterset WE8MSWIN1252. I know that schemas and objects will be automatically created in 11g.
3) However, the problem I face is that this importing of data has been running for the past 4 hours and counting and it is still not done.
My question is:
Is it because of the difference in the characterset in the dmp file and 11g that is causing this importing to be slow? Could it be that characterset conversion is causing high overhead?
OR
Is it because I exported all the indexes from 9i and now during importing, it is taking a long time to create those indexes in 11g? Should I have export full but set index=F and create a index creation script from 9i and run it in 11g, so as to save time?
OR
Both of the above is causing the importing to keep on running? Or ??
Edited by: moslee on Nov 21, 2012 11:54 PM
Edited by: moslee on Nov 22, 2012 12:01 AM