Allow Optional CHAR-to-VARCHAR Conversion in DataGate to Preserve Application Transparency between Db2 for z/OS and Db2 LUW

When replicating data from Db2 for z/OS (EBCDIC) to Db2 LUW (UTF-8), DataGate increases the length of CHAR columns to accommodate Unicode expansion. While this behavior is understandable from a data integrity perspective, it can introduce application compatibility issues.

A common use case is flag columns defined as:

FLAG CHAR(1)

Containing values such as:

'S' = Yes,
'N' = No

After replication, DataGate may create the target column as CHAR(2), causing values to be returned as:

'S ',
'N '

Instead of:

'S',
'N'

This difference can lead to application failures when programs perform exact value comparisons. It also forces applications to use functions such as RTRIM(), which may negatively impact index usage and query performance.

We would like DataGate to provide an optional configuration to map source CHAR columns to VARCHAR columns in the target Db2 LUW environment, preserving the logical value while avoiding trailing blank issues.

Ideally, this configuration should be available at the table level (or even column level), allowing customers to selectively enable the behavior only for specific tables where it makes sense, rather than applying a system-wide configuration.

Benefits:

• Improved application transparency between Db2 for z/OS and Db2 LUW.

• Reduced need for application code changes.

• Elimination of issues caused by trailing blanks in flag and code columns.

• Better preservation of index efficiency by avoiding the need for functions such as RTRIM() in predicates.

• More flexibility for customers with mixed workloads and different data requirements.

This enhancement would be particularly valuable for environments migrating or offloading workloads from Db2 for z/OS while maintaining compatibility with existing applications.