on 01-31-2012 5:43 AM
Dear all,
We are doing import/export in our landscape
We successfully completed Development n quality and also training system
but we are facing problem in Production system (ie Export is successful in 10 - 12 hours)
but Import is taking Days till now we hadn't completed( we are testing The PRD data in Trail system)
System details
Database size is 3.5 TB
Ecc 6 with oracle 10g in suse linux 10 (source system)
Target
Ecc 6 with oracle 11g in suse linux 11
Kindly help us where we are wrong
is any parameters need to be set
Babu
If, and only if you are waiting on disk io and you are using lvm, then you may need to set barrier=0 in the filesystem mount options. We are doing a migration on a SLES 11 system and the import (24 R3load processes) was speed up by factor 5.
I/O Performance difference between SLES10 and SLES11 on ext3 mounted lvm devices
Cheers Michael
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
do you know what particular R3load process is taking so long?
Is it table import, index creation?
you can check runtime with import_monitor.sh tool from MIGTIME package.
Have you adatped ORACLE profile for import as per note Note 936441 - Oracle settings for R3load based system copy
Regards
Stanislav
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Some more details
Our Envirmoment as follows
SAP details
SAP ECC6.0, Oracle 10.2.0.5, SuSE10 SP3
Database size - 4TB
SAP Kernel 700 Patch 291
Hardware details
IBM x3850 X5 (64 Core CPU & 512GB RAM)
SAP ECC6.0, Oracle 11.2.0.2, SuSE11 SP1
SAP Kernel 700 Patch 29
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
To speedup the export/import process you can think of the following options
1) Table splitting : Split larger tables into smaller chunks of multiple sizes
2) Package splitting: Splitting SAP export packages into multiple chunks of smaller sizes
3) Increase R3load process: Increase the number of R3load processes until CPU utilization reaches around 80%
4) Use Distribution monitor to utilize CPU resources available on Application server. You can add up R3load processes on Application server.
5) Ensure that System statistics, Oracle Dictionary statistics and missing statistics are updated before performing export/import.
6) Increase the size of Redo log files to 250 GB.
7) Increase the number of redo log group members to atleast 16.
Ensure that above recommendations are implemented during the migration project. If not, you will land up in issues like export and import taking more time.
In case you have already considered above recommendations, please check alert_<SID>.log file for more details. Monitor log swicth operation in the log file.
Additionally monitor CPU usage and increase R3load processes accordingly.
Regards,
Deepak Kori
Hi deepak
h16) Increase the size of Redo log files to 250 GB.
h17) Increase the number of redo log group members to atleast 16.
is it necessary to increase the size of redolog file to 250 Gb??? ,
we kept as 1 gb
in our scenario
export has completed in 8 hrs
but import is taking too much time like 50- 60 hrs
is it necessary to increase the size of redolog file to 250 Gb???
It may be necessary to increase size if (and only if) you can see a lot of Checkpoint not complete messages in your Oracle alert log file.
So first have a look there!
By the way, 250GB will be oversized imho.
Edited by: Joe Bo on Feb 1, 2012 7:56 AM
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.