Limitations

Consider the following limitations if you are planning to use the Automatic Schema Evolution functionality in the preview release:

. Parallel Replicat is not supported in the preview release.

  1. MAPEXCEPTION clause is not supported with AUTOSCHEMA. You must remove the MAPEXCEPTION clause to use AUTOSCHEMA.

    However, if the MAPEXCEPTION table is created previously, then no error is thrown and the process will continue.

  2. Automatic Schema Evolution does not attempt to propagate the hidden, invisible, or pseudo columns onto the target tables. One exception on the pseudo column(s) is that if the pseudo column is part of the source table’s key, then it is included in the target table created by AUTOSCHEMA.

  3. Automatic Schema Evolution currently does not guarantee the same key column order as used in the source table.

    For example, on the source, if the table is created using:

    CREATE TABLE
    QASOURCE.SRCTAB(COL1 INT, COL2 VARCHAR2(100), COL3 INT,
    PRIMARYKEY(COL3, COL1);

    Automatic Schema Evolution does not guarantee generating the DDL with the PRIMARY KEY clause with COL1 and COL3 in the same order as defined on the source table. It can generate PRIMARY KEY (COL1, COL3)), as it entirely depends on the order of columns written in the trail table definition record (TDR).

    See Support for Automatic Primary Key and KEYCOLS for details.

  4. Clauses such as auto_increment are not supported with AUTOSCHEMA while creating the target table.

  5. AUTOSCHEMA does not replicate isolated check constraint clause or any additional indices that exist on the source table(s).

  6. COLMAP support is not available with DROP COLUMN.

  7. COLMAP is not processed with ALTER TABLE.

  8. The table definition record (TDR) in the trail file does not contain enough information on the date-time precision of multiple components of a source interval type. For example, if the source interval column is defined as INTERVAL DAY(x) to SECOND(y), then TDR does not contain information about x and y.

Tthe source interval type is mapped to an equivalent interval type on the target if it supports corresponding interval type and if it can accommodate the maximum size of the source interval column. Otherwise, source interval column is mapped to VARCHAR type on the target.

  1. IOT/Blockchain tables are created as normal tables for Oracle and non-Oracle targets. If you want to create blockchain tables for Oracle targets then you can use DDL replication because AUTOSCHEMA is disabled for Oracle-Oracle replication.

  2. Signed/Unsigned decimal/Float/Double from MySQL source is created as signed counterparts on target databases.

    Unsigned decimal from MySQL is being mapped to signed decimal on target as target does not support decimal as unsigned.

    Unsigned decimal/double/float data type from MySQL source gets created as signed decimal data type on target in mysql-mysql replication as MySQL trail format currently does not provide information about the signed/unsigned decimal/floats/doubles.

  3. Oracle data types, rowid and urowid cannot be differentiated using the fields in the TDR record.

  4. Only Classic and Coordinated Replicat types are supported in this release. Parallel Replicat is not supported for Autoschema.

  5. ALTER TABLE is only suppported only for ADD COLUMN in the preview release.

  6. Column level charset is not supported.

  7. Automatic Schema Evolution is incompatible with COLMATCH.

  8. If the ASSUMETARGETDEFS OVERRIDE is enabled in the Replicat parameter file, and if the corresponding object does not exist in the target database, Replicat abends with the table does not exist error, even when AUTOSCHEMA CREATETABLE is enabled.

  9. If the SOURCEDEFS OVERRIDE is enabled in the Replicat parameter file with AUTOSCHEMA also enabled, and if the corresponding object does not exist in the target database, Replicat would abend with the error stating that the DDL replication cannot be used on the table with SOURCEDEFS.

  10. Only the primary key information or one unique key information in the absence of primary key is replicated to the target. Additional unique keys or any indices created at the source are not replicated into the target. Similarly, Foreign Key information is not replicated into the target.

  11. Partition Tables are not supported in Automatic Schema Evolution.