Preparing data model files for metadata import (IBM Knowledge Catalog)

Data models depict the structure of a data store, how its various data elements are related, and the attributes that are associated with each element. You create such visual representations by using a data modeling tool. You can then export such diagram from the data modeling tool and use metadata import it to add the model to a catalog to have a single collection point for information about the database design and your data assets. Data models in the catalog are read-only copies of the originals created and maintained in database modeling tools.

Prerequisites

Before you can view a logical or physical data model in a catalog in IBM Knowledge Catalog, you must complete some prerequisite steps and make sure certain conditions are met:

Export the data model from the data modeling tool. For each data modeling tool, the exported file must have a specific format so that the data model can be properly added to a catalog.

Data modeling tool	Required export format
ER/Studio	.dm1
erwin Data Modeler	standard .xml
SAP PowerDesigner	.ldm or .pdm

Data models can be large. If you have a particularly large data model, exclude any diagrams before you export the model. Diagrams just add to the size but are not processed by the import. Also, consider storing each data model in a different catalog for the following reasons:

Cascading deletes are not available for these assets in IBM Knowledge Catalog. You must delete all assets individually. Keeping data models in separate catalogs makes deleting the full data model easier, without impacting other data models.
Model attributes are imported as separate assets to increase their visibility and make them more flexible to use in IBM Knowledge Catalog. Therefore, large data models can have hundreds or even thousands of assets in the catalog. Separate catalogs help managing such large data models.

Compress the exported data model file before you upload it to a project in IBM Knowledge Catalog for further processing. The compressed data model file must have .zip format. When you create the metadata import, this .zip file constitutes the data scope.

When you import a physical data model, the .zip file must also include a connections.ini file with the following content for each model:
```
[<physical_model_name>]
Connection_String=<connection_string>
Type=<connection_type>
Server_Name=<server_name>
Database_Name=<database_name>
Schema_Name=<schema_name>
User_Name=<user_name>
```
The name of the physical model, Connection_String, and Type are mandatory elements. The other elements are optional. If you don't need an element, omit it.

The Connection_String value is either a dictionary identifier (each connection in MANTA Automated Data Lineage must contain a unique dictionary identifier) or a connection string that points to a specific technology. Type is case-sensitive and can have the following values:
- BigQuery
- DB2
- Hive
- JDBC
- Kafka
- MSSQL
- Netezza
- ODBC
- Oracle
- PostgreSQL
- SAPHana
- Snowflake
- SSAS
- Sybase
- Teradata
- Unknown
In MANTA Automated Data Lineage, the technology of the source system is mapped based on the specified connector. Specific connectors like Hive, Oracle, and Teradata can be mapped directly. For generic connectors such as ODBC, JDBC, and Unknown, all technologies are searched to find a suitable source system. The first technology (the order is nondeterministic) that returns a suitable source system is used.
After you upload the .zip file to your project, create a metadata import and select this compressed data model file as import scope. You can select one data model file per import from your project. After you set the scope, you must also make the originating data modeling tool known to IBM Knowledge Catalog. The metadata import service relies on this information for evaluation of the file format and structure after unpacking the .zip file. Metadata import then creates data model assets in the selected catalog based on the content of the data model file.