SAP Power Designer forward engineering to SAP HANA

[Former Member]

Hi Community,

In my current project we are building an EDWH architecture with SAP HANA in the center. SAP Power Designer is used as tool to design the physical data models. We want to govern the entire development process via Power Designer, hence every table should be first defined there. In the end we want to forward engineer the model to SAP HANA, preferably using the feature "Apply Model Changes to HANA Repository" in Power Designer. However when trying to do this we are getting the following issue:

After going back and forward again in the dialog, we receive another error:

I have two questions:

1. Does anyone have an idea what causes these errors and how to resolve them?
2. Will this feature assign the tables to a package in the content as well?

We are using the following versions:

• Power Designer Client, Version 16.5.5.4
• SAP HANA SPS 10

Kind Regards

Matthias

[former_member185317]

Take a look at the specific Help documentation on using PowerDesigner with SAP HANA objects.

http://help.sap.com/saphelp_pd1655_data/helpdata/en/c7/c20a926e1b1014b72bd8da3188e85a/content.htm

Generating all of the HANA objects.

http://help.sap.com/saphelp_pd1655_data/helpdata/en/c8/16819f6e1b10149bfdd0d2fd9ef4e2/content.htm

HTH,

-Matt C.

[c_baker]

Before you can apply the model changes to the database, the objects must exist in the HANA repository, just like any other PD 'update existing' model capability.

The steps you need are as follows:

1. You must first create the tables using normal DDL or ODBC in the HANA server.  For DDL, you can apply it using hdbsql or HANA Studio (using a console opened in the schema).
2. When you have done that, the thing to also understand is that the HANA repository will also not show any tables or objects for applying changes to, unless you also have an analytics object defined under a package in your model (and eventually in HANA.  This is needed for applying the changes (the underlying tables are what will drive the analytic object, so tables will then show as selectable for the 'Apply Changes').  In other words, the HANA repository will not show anything unless an analytic object (OLAP cube) is defined in you model.
3. Now in your model, create a simple multidimensional OLAP cube as well.  The best way to do this is to define your fact and dimension attributes on some tables and then use the 'Tools/Multidimensional Objects/Generate Cube ...' wizard.  The package you create with the OLAP cube will eventually show in the 'Contents' folder under your HANA Studio login.
4. To apply changes, you also need to have an archive model for PD to perform the comparison of your changes to the 'production' version in HANA, so make sure once you generate the DDL, you also save the .apm file.  (You can prepare for using it by simply starting the 'Database/Apply Model Changes to Database' dialog, selecting the .apm file and then press 'Apply' and 'Cancel').
5. Now you can run the 'Database/Apply Model Changes to HANA Repository' dialog.  As you now have an analytic object in you model, You can push all tables and analytic objects to the HANA repository for round-trip engineering.  Even though you may only have a couple of tables used by the analytic/OLAP object, ALL tables can be selected for update/import into the repository in the wizard as it runs.

One thing to remember is that the default creation of an OLAP cube in PD will use some of the same attribute and measure names in the facts and dimensions.  This is fine in the PD namespace, but in HANA, all names must be unique within the package.  This will not be picked up in the normal check model functionality.

I have a HANA extension that can help here, or you can ensure all names are unique yourself.

This was actually part of a workshop I gave at TechEd 2013 in Las Vegas (RDP275) which also covered replication from ASE to HANA (now the subject of our A4A offering).

Chris