Known issues and limitations for Data Gate

Limitations

Source and target limitations

You can use Data Gate to synchronize a single data source (Db2 for z/OS subsystem or data sharing group) with exactly one target database instance (Db2 or Db2 Warehouse)

• To connect multiple data sources and target database instances, you must provision multiple Data Gate instances, that is, one for each data source/target combination.
• You can connect multiple Data Gate instances to the same data source in order to implement high availability (HA) and disaster recovery (DR) architectures. For details, see High-availability.
• You can use a single database instance as the target of exactly one Data Gate instance. That is, you cannot make one and the same database instance the target of multiple Data Gate instances.

Columns with a data type of XML or LOB are not supported

Columns with a data type of XML or LOB are not supported. A Db2 for z/OS table with unsupported columns can be added into Data Gate, then be loaded and synchronized. But the values of the columns with these unsupported types will therefore be missing in the target table.

Table load and synchronization conditions

A table cannot be loaded and synchronized if one or more of the following conditions apply:

• A mask has been defined for the table with a CREATE MASK statement.
• A permission has been granted on a table column with a CREATE PERMISSION statement.
• One or more table columns have a security label.
• One or more table columns use a field procedure (FIELDPROC).

Important: If you try to load and synchronize a table that satisfies one of these conditions, the Data Gate server on Cloud Pak for Data stops and then restarts automatically after an outage period of five to ten minutes. Hence, you need not do anything to recover from the server shutdown.

Indexes and access control lists are not inherited

Indexes and access control lists defined on the Db2 for z/OS source tables will not be inherited by the Data Gate target tables.

Different treatment of trailing blanks when row store is used in target database

SQL queries against the target database (Db2 LUW row store) lead to result values that are different from those obtained by running the same query against the source database (Db2 for z/OS). The difference occurs because a Db2 LUW column store cuts off the trailing blanks to normalize the values, reducing them to the semantically relevant part. In a Db2 LUW row store, however, the trailing blanks are kept, and that leads to different query results for some queries. This issue surfaces only if users have configured a Db2 LUW target database as a row store, and if they use Data Gate's query acceleration feature. Furthermore, the problem seems to occur only if the query contains a UNION, DISTINCT, or GROUP BY clause.