BW on HANA scaleout Migration Part 1: Preparation Checklist and known issues
I recently worked on a BW on HANA Scale out Migration Project and in this Blog I would like to describe the preparation that was done for the Migration and in particular problems that came up that you can avoid in your Projects. In Part 2 I will cover the Migration itself and the issues I encountered. In Part 3 I will cover the post migration activities and any problems that we needed to resolve.
A good starting point that’s gives a good overview of the complete process for a BW on HANA scaleout migration is Marc Hartz’s Best practice document that you can find here: http://scn.sap.com/docs/DOC-39682. This document is constantly updated and is a must read before starting any BW on HANA Scale out Migration project.
Before starting the Migration you need to ensure that the following is done:
1: You have the Latest HANA Revision installed on the HANA DB.
The Import may use a lot of log space so it is highly recommended to set the log mode to overwrite on the HANA DB before starting the Import to avoid a log full scenario during the import. Please do not forget to restart the HANA DB after changing the log mode, this is important!! Where to find the parameter is shown below:
In the HANA Studio look under TAB Configuration>Global.ini>:
Afterwards when the Migration is finished you can set the log mode to normal again if you want to have log backups of the system.
2a: You are using the latest BW media for the Import. You can find the BW Media here: SAP Service Market Place The path to the BW Media is: Installations and upgrades > Browse our Download Catalog > SAP NetWeaver and complementary products> SAP NetWeaver BW powered by SAP HANA> SAP NetWeaver BW 7.3 powered by SAP HANA> BW ON HANA INSTALLATION 1.0. From here you need to download the relevant Kernel file, the SL Controller file and the latest SAP HANA Platform Edition files. Store the BW media and the software you downloaded on the application server where you will install the BW system.
2b: To do the Migration I used the Software Provisioning Manager tool (at the time SP 01 Patch 02 was the latest version). This is intended to replace sapinst and if you use this tool for the migration then you don’t need to do a lot of manual configuration like described in the note https://service.sap.com/sap/support/notes/1715048 You can find more information on the software provisioning manager in the user guide: Software Provisioning Manager With the software provisioning manager it will by default create the installation directory in the tmp directory (/tmp in our environment) so before starting you need to make sure there is enough space in tmp directory. The documentation says at least 600m but from my experience you need about 1- 2G to be comfortable. If you need to start the installation again then the software provisioning manager will create a new install directory every time. You can delete the old install directories to free up space. The other option is that you change the tmp directory to some other directory with more space before starting the installation. How to change the default directory for SWPM is explained on page 61 under the heading Useful Information About the Installer. Before doing the export or the Import using the software provisioning manager please read the note https://service.sap.com/sap/support/notes/1775293 and the note https://service.sap.com/sap/support/notes/1706930 for latest known problems and available work around’s.
3: For the export that you use for the DB Migration it is critical that the export is done using the latest Kernel, DBSL and R3 Load version and patches and also the latest BW support package when you are working on a BW scale out scenario. For large tables it is really important that good table splitting is done for the performance of the export and the Import. Some information I did not know at the time of the migration is that unlike the 720 kernel the 721_EXT Kernel contains the so called MASSLOADER functionality which improves the load performance with LOB data. To enable it you have to set the Environment variable HDB_MASSIMPORT=YES. So you should use this Kernel release for export and Import if possible to improve the performance.
4: To ensure that the Tables were distributed correctly on the nodes during the Import on the HANA scale out DB we did the customizing described in the SAP note https://service.sap.com/sap/support/notes/1819123 (If your HANA revision is 63 or higher please follow the steps in the note 1908075 instead). In short there were two steps required in our case:
As part of the export execute the report SMIGR_CREATE_DDL and activate all SAP HANA-specific options, in particular the option “Row Store List”. With this option a file ESTIMATED_ROW_COUNT.TXT is generated that contains the estimated number of rows of the BW tables. Copy this file to the /ABAP/DB directory of SWPM. As of SWPM 1.0 Support Package 2 this file is copied automatically.
The file TABLE_PLACEMENT.TXT is attached to the SAP Note https://service.sap.com/sap/support/notes/1819123 Download this file and save it to a directory that can be accessed by SWPM during the Import. In the dialog phase of SWPM for SAP HANA landscape reorg parameters, choose the option “Use a parameter file” and enter the complete path including the file name as the “Parameter File”.
During the Import you will come to this screen where you can point to the file:
5: Before starting the Import ensure that the BW application server and the HANA DB have the same time zone setting, the note https://service.sap.com/sap/support/notes/1706930 has further information on this.
6: Another useful source that I was pointed to after the migration which applies not just to BW but to all Netweaver ABAP stacks is the “End-to-End + Best Practice Guide – Database Migration to SAP NetWeaver Application Server ABAP on SAP HANA”, you can find the guide here: