Setting isolation levels for non-Sybase replicate data servers

Isolation levels may vary depending on the replicate data server. This has an impact on configuring parallel DSI in Replication Server.

• Oracle – READ COMMITTED and SERIALIZABLE

• Microsoft SQL Server – READ UNCOMMITTED, READ COMMITTED, REPEATABLE READ, SNAPSHOT, and SERIALIZABLE

• IBM DB2 UDB – REPEATABLE READ, READ STABILITY, CURSOR STABILITY, and UNCOMMITED READ

The rs_set_isolation_level function string must be edited for non-Sybase replicate data servers, and include the rs_isolation_level system-defined variable. See the Replication Server Reference Manual for more information about rs_set_isolation_level.

If you are using a data server other than Adaptive Server, make sure you include the rs_isolation_level variable when you modify the rs_set_isolation_level function string for your data server.

To set an isolation level, create a function string in the appropriate function-string class. For example, in:

• Oracle – to set the SERIALIZABLE isolation level:

  create function string rs_set_isolation_level
  for rs_oracle_function_class
  output language
  ‘set transaction isolation level serializable’
  

• Microsoft SQL Server – to set the SERIALIZABLE isolation level:

  create function string rs_set_isolation_level
  for rs_msss_function_class
  output language
  ‘set transaction isolation level serializable
  

• IBM DB2 UDB – to set the REPEATABLE READ isolation level:

  create function string rs_set_isolation_level
  for rs_udb_function_class
  output language
  ‘set current isolation = RR’