Enterprise Replication Enhancements

Enterprise Replication conversion and reversion is now manual instead of automatic. For instructions, see the IBM Informix Migration Guide. For the error messages, see the IBM Informix Administrator's Reference.

Dynamic Server, Version 9.3, includes extensibility enhancements, performance improvements, functionality enhancements, and command-line changes for Enterprise Replication.

Replication of Extensible Data Types

Enterprise Replication provides support for replicating the following extensible data types:

• Data stored as smart large objects in sbspaces (CLOB and BLOB data types), CLOB and BLOB columns (explicitly specified in the table schema), and updates to CLOB and BLOB columns (with some restrictions)
• Opaque user-defined types (UDTs)
• Multirepresentational data types, if the required stream support functions exist.

  For information on writing the required functions, refer to the IBM Informix DataBlade API Function Reference.

• IBM Informix Spatial DataBlade module

Version 9.3 does not include support for replication of the following user-defined types:

• Row types
• Collections
• Lists
• Sets and Multisets

Enterprise Replication allows the following (with some restrictions):

• UDT column references and UDRs in replicate WHERE clauses
• UDTs for primary key columns

For more information, see IBM Informix Dynamic Server Enterprise Replication Guide:

• Replicating simple and smart large objects
• Considerations for replicating opaque data types
• UDT support functions

Support Functions for Replication of User-Defined Types

To replicate UDTs, Enterprise Replication requires that the UDT designer provide two support functions: streamwrite() and streamread(). The streamwrite() function converts the UDT column data from the in-server representation to a representation that can be shipped over the network. On the target server, Enterprise Replication calls the streamread() function for each UDT column that it transmitted using the streamwrite() function.

For more information, see the section on writing opaque-type support functions in the IBM Informix DataBlade API Programmer's Guide.

Performance Enhancements to Enterprise Replication

Enterprise Replication includes the following performance improvements to parallel processing:

• Enterprise Replication now applies all replicates (in replicate sets and individually) in parallel by default.
• Enterprise Replication threads now apply transactions from the same source in parallel unless they contain updates to the same row.
• Enterprise Replication threads normally commit on the target in the same order as on the source.
• Enterprise Replication threads can commit out of order on the target if there are no conflicts.
• Enterprise Replication now uses buffered logging to apply transactions.

Improved parallel processing is built in and requires no user configuration or interaction. However, this feature is automatically disabled if you are using page-level locking.

SERIAL Column Primary Keys

The CDR_SERIAL configuration parameter enables control over generating values for serial and SERIAL8 columns in tables defined for replication. This feature is useful for generating serial column primary keys in an Enterprise Replication environment.

For more information, see CDR_SERIAL in the section on configuration parameters in the IBM Informix Dynamic Server Enterprise Replication Guide.

Replicate Sets and Exclusive Replicate Sets

You can manage replicates individually and as part of a replicate set. Put tables in exclusive replicate sets to guarantee referential integrity between tables when you use any form of time-based replication.

For more information, see creating and managing replicate sets in the IBM Informix Dynamic Server Enterprise Replication Guide and migrating Enterprise Replication data in the IBM Informix Migration Guide.

Replicating Only Changed Columns

Enterprise Replication provides the ability to replicate only the changed columns, rather than the entire row.

If only changed columns are replicated, the data for all replicated columns might not be available for spooling to the ATS (Aborted Transaction Spooling) and RIS (Row Information Spooling) files. Therefore, the format for these files has changed.

For more information, see the IBM Informix Dynamic Server Enterprise Replication Guide regarding:

• Replicating only changed columns
• Aborted Transaction Spooling files
• Row Information Spooling files

Spooling of Replicate Data to Nonlogging Smart Large Objects

Enterprise Replication spools row data in the send and receive queues to an sbspace that you specify in the CDR_QDATA_SBSPACE configuration parameter. You can control logging of these sbspaces.

Enterprise Replication spools transaction records from the send and receive queues to a dbspace that you specify in the cdr_qhdr_dbspace parameter.

For more information, see the IBM Informix Dynamic Server Enterprise Replication Guide regarding:

• Setting up send and receive queue spool areas
• Defining replication servers
• Specifying CDR_QDATA_SBSPACE and CDR_QHDR_DBSPACE configuration parameters

In-Place Alters to Add or Drop Shadow Columns (CRCOLS)

Enterprise Replication uses shadow columns for conflict resolution. The database server now processes the following ALTER statements for adding and dropping shadow columns as in-place alters in most cases:

ALTER TABLE ... ADD CRCOLS
ALTER TABLE ... DROP CRCOLS

In-place alters are quick because the database server updates each row in place instead of copying the entire table. The in-place processing of these ALTER statements requires no user action.

New onstat Options for Enterprise Replication

Use the following onstat options to obtain information about replication of user-defined routines (UDRs):

• onstat -g dss UDR
• onstat -g dss UDRx
• onstat -g grp UDR
• onstat -g grp UDRx

For details, see the appendix on onstat commands in the IBM Informix Dynamic Server Enterprise Replication Guide.

The cdr finderr Utility

This release includes updates to the command-line interface to support new features, including a new cdr finderr utility which looks up a specific Enterprise Replication error number and displays the corresponding error text.

For more information, see the command-line utility reference in the IBM Informix Dynamic Server Enterprise Replication Guide.