The default conversion procedure is to export the entire database using SAPinst (which controls R3load), create a new Unicode database, and then import the database using SAPinst again (internally R3load). You can choose either the “Two Server Method” which is recommended. In this case you build the Unicode database on an additional server. If you choose the “One Server Method” you must delete the non-Unicode database before. In fact you perform both a system copy using SAPinst (as described in the relevant System Copy Guide mentioned below) and simultaneously a database conversion to Unicode. The conversion is performed by R3load during the export procedure. For building the Unicode system you use the export files generated during the export procedure instead of the export media provided by SAP.
Note
You can choose either the homogeneous or the heterogeneous system copy method. If you want to perform a database or platform change during the conversion process, choose the heterogeneous system copy method.
Documentation:
For this chapter you must use the System Copy Guide in parallel. You can download the valid System Copy Guide for your release from www.service.sap.com/instguides.
Downtime Estimation
If you want to gain a rough estimation of the expected system downtime, run report UMG_R3LOAD_RUNTIME_PREDICTION in transaction SE38.
In addition SAP Note 857081 provides information about methods and tools for optimization of the export and import procedure.
Do not delete non-Unicode system immediately after the import
After both export/conversion and system copy has been successfully completed, SAP recommends keeping the non-Unicode system for the duration of the testing period of the Unicode system. Afterwards the non-Unicode system can be deleted using the standard procedure.
IBM DB2 UDB for UNIX and Windows: Tablespace sizes for conversion Unicode - non Unicode
To get the appropriate tablespace sizes for IBM DB2 UDB for UNIX and Windows, install R3szchk BEFORE the conversion. You can download from SAP Service Marketplace at service.sap.com/patches <your product> <your release> → Binary Patches. To avoid wasting space in your DB2 UDB for UNIX and Windows tablespaces we strongly recommend that you create all tablespaces in the target database with extentsize 2.
Note:
This is not the default for all SAPinst releases up to and including SAP NetWeaver 2004 SR1.
You therefore need to manually adjust the extentsize in the SAPinst tablespace dialogs. The pagesize for a unicode system should be 16K as proposed by SAPinst.
3.1 Database Export and Import
The following section describes the R3load copy procedure of non-Unicode system to Unicode system.
Make sure that all required CDs/DVDs are available now as described in section Software Requirements in this document [page 9].
Import the latest SAP Kernel patch as described in SAP Note 19466 and the applicable R3load version.
Requirements: Use the latest R3load version and the latest version of R3ldctl and R3szchk for your release
Documentation: - this document
- System Copy Guide for your release. The System Copy Guides can be downloaded from:
http://www.service.sap.com/instguidesNW70
For detailed information about optimization of the system copy procedure go to https://service.sap.com/systemcopy -> optimization.
Status: required.
Perform the export from the non-Unicode system (source system), install the target system and import the database files that you have exported from the source system as described in the System Copy Guide:
“Homogeneous and Heterogeneous System Copy for SAP Systems Based on <RELEASE>”
Section: System Copy Procedures on <OPERATING SYSTEM>”
It is not possible to perform the Unicode conversion during the import! Make sure you convert the MDMP data during export phase by calling R3load with parameter '-datacodepage 4102' or '-datacodepage 4103' (details see section 3.1.1 Database Export).
Read section Software Requirements [page 9].
Unicode Conversion with database change: Important information for Informix databases
If you plan to convert an Informix database, follow the instructions in SAP Note 811431 BEFORE the database export!
IBM DB2 Universal Database for iSeries:
Before starting the installation, follow the instructions in SAP Note 1158503 “iSeries: Setting environment Variables for SAPinst”.
3.1.1 Database Export
1. Perform the necessary preparation steps before starting SAPinst.
2. Read the document “Unicode Conversion: Tips and Tricks for Improving the Conversion Time”.
3. Perform the database export in the non-Unicode system (source system) as described in the relevant section of the System Copy Guide: Running SAPinst to Export the Database.
4. When running SAPinst to export the database and SAPinst prompts for General Export Parameters Data file code page make sure the code page of the target system depending on the byte order of its CPU is entered (see SAP Note 552464):
If the target platform (e.g. Linux, OSF1, NT) uses a little endian CPU (e.g. Alpha or Intel CPU), enter 4103.
If the target platform (e.g. HP-UX, AIX, SunOS, OS/390) uses a big endian CPU (e.g. Sun SuperSPARC), enter 4102.
Caution
Be aware that this parameter depends on the processor architecture, not on the operating system.
Make sure that the most recent non-Unicode R3ldctl (SAP Kernel) is used for the export.
Note the additional information for the Unicode Conversion during the export of the database:
Server Necessary Steps Further Information
MS SQL Server only During the installation, the collation setting must be fixed.
See SAP Note 600027
Oracle only When running SAPinst to export the database and SAPinst prompts for Database Instance Parameters General Settings Database character set make sure the character set of the source database is entered (e.g.
UTF-8, WE8DEC or US7ASCII).
See System Copy Guide, sections Input for Exporting Your Database
All servers During the export of the source system R3load writes log files into the installation directory of the source system. The file names are like SAP*.log for example SAPAPPL2.log. If you create own packages for the files, the file names are like
<TABLENAME>*.log.
The SAP*log files or
<TABLENAME>*.log files should be saved for later error analyis.