Determines whether Sybase Replication Agent converts non-Sybase temporal datatypes to the Sybase datetime format.
true
true – Sybase Replication Agent converts all data in the primary database native date/time datatypes to the Sybase datetime format.
false – Sybase Replication Agent replicates data in the primary database native datetime datatypes as character strings using the Sybase rs_oracle_datetime datatype.
The pdb_convert_datetime parameter is provided for backward compatibility with previous versions of Sybase Replication Agents and Replication Server. If you use Replication Server version 12.0 or later, Sybase recommends that you use the Replication Server heterogeneous datatype support (HDS) feature for all datatype conversion and translation.
Sybase Replication Agent checks the value of the pdb_convert_datetime parameter at the time an object is marked for replication. Transaction log objects that support replication of the marked object are constructed to provide the desired date format.
If you change the value of the pdb_convert_datetime parameter after an object is marked, it has no effect on the marked object. To change the datetime datatype conversion for a marked object, you must unmark the object, change the value of the pdb_convert_datetime parameter, then re-mark the object.
For trigger-based Replication Agents, the actual conversion of datatypes takes place when an operation is recorded in the Replication Agent transaction log. For log-based Replication Agents, the conversion takes place after the log records have been read and before LTL is generated to send to the Replication Server.
Any missing component in the non-Sybase date/time datatype format is treated as an implied 0 (zero) when it is converted to the Sybase datetime format.
When the value of the pdb_convert_datetime parameter is true, the replication definition for each table should specify that the declared datatype for all date/time columns is datetime.
If the value of the pdb_convert_datetime parameter is false, the Replication Agent sends date/time data to the primary Replication Server as default-sized character strings. The default character string size varies by database and datatype:
DB2 Universal Database: DATE = char(10), TIME = char(8), TIMESTAMP = char(16)
Microsoft SQL Server: datetime or smalldatetime = char(23), timestamp = binary(8)
Oracle: DATE = char(8)
If the non-Sybase date/time format requires a longer string to be replicated correctly, you can either set the value of the pdb_convert_datetime parameter to true, or modify the trigger-based Replication Agent table marking script (mark.sql) to create larger shadow table date/time columns.
Set the value of the pdb_convert_datetime parameter to true if all date/time values replicated from the primary database will be replicated as the Sybase datetime datatype.
pdb_convert_datetime must be false if a table containing replicated LOB columns has datetime datatype in the primary key.
Replication Agent date/time datatype conversion does not work with LOB column replication, unless either of the following conditions exist (these conditions are not required for Oracle):
There are no date/time columns in the tables that have LOB column replication enabled, or
The primary keys in tables that have LOB column replication enabled do not contain date/time datatypes.
Otherwise, if you use the pdb_setrepcol command to enable LOB column replication, you must set the value of the pdb_convert_datetime parameter to false.
The Replication Agents, other than Oracle, query LOB data directly from the primary database (the LOB data is not captured by the triggers and is not logged for UDB). To successfully query a primary database table for a LOB column value, any date column value must retain the primary database format and structure for the date value to appear correctly in the query. The format and structure for the date value cannot be converted to the Sybase datetime format.
Sybase recommends that you set the value of the pdb_convert_datetime parameter to false for better Sybase Replication Agent throughput performance and optimal datatype handling.
