Along with today's announcement about the new version of DB2 and InfoSphere Warehouse, IBM also announced an updated release of InfoSphere Data Replication to support the new DB2. All of IBM's data replication technologies - CDC, SQL Replication, and Q Replication - have been updated to support log capturing from DB2 10 as well as applying of changed data to it. However, Q Replication has also been extended to improve DB2 and InfoSphere Warehouse availability solutions.
Key enhancements in support of DB2 LUW 10 include:
- The ability to create schema-level subscriptions in Q Replication.
- Give Q Replication a pattern, such as tolleson.*, and subscriptions will be created automatically when a newly created table matches that pattern.
- The DDL will also be replayed on the target system and data replicated to it.
- This reduces the TCO of data replication.
- The expansion of Q Replication's automatic DDL replication to include all of the following:
- Create and drop table, which can be done for both bidirectional and unidirectional configurations. This replicates complete table creation with indexes, triggers, and more added prior to first table usage. It also includes initial load of the new table at the target, if required.
- Alter add column
- Alter data type
- The option to suppress replication of trigger-generated rows:
- If you have matching triggers on source and target, you can tell Q Capture not to replicate rows that are create or changed as a result of a trigger firing. This is a great way to increase throughput by reducing both the amount of data shipped and also amount of work Q Apply has to do. That includes dealing with conflicts that would result from the trigger firing on the target just before the source's trigger-generated rows are applied :)
- The option to suppress replication of child rows affected by cascade delete and set null:
- If you have matching RI (referential integrity) on source and target, you can tell not to replicated the deleted or changed rows for for dependent tables affected by Delete Cascade and Set Null. Like suppression of trigger-generated rows, the child row options reduce the amount of data shipped and the amount of work Q Apply has to do, including conflict handling.
A few things might not be obvious from the DB2 announce:
- DB2 Advanced ESE and InfoSphere Warehouse now include three-site Q Replication at no additional cost. This adds one site to what was available in 9.7. This provides a lower cost solution for multiple secondaries or having a time-delayed standby along with an active-active configuration.
- Not everything in the Control Center was deprecated :) Specifically, the Replication Center was separated from the Control Center and is provided as a stand-alone tool with DB2 and InfoSphere Warehouse v10. The Replication Center supports both Q Replication and SQL Replication.
A few related points that are probably not obvious from the announcement for IBM's InfoSphere Data Replication (IIDR):
- 9.7 is the last release of InfoSphere Replication Server on UNIX and Windows. If you were current on support, IBM automatically upgraded you to IIDR.
- All of InfoSphere Replication Server 9.7's function is now found in IIDR 10.1.
- Starting with IIDR v10.1.3, all new Q and SQL Replication function is found only in IIDR. That includes everything mentioned in this post.
- The Q Capture in DB2 10 can read log records from DB2 10, but no DB2 9.7 and older. The opposite is true for the 9.7 Q Capture. It cannot read log records from DB2 10, but can definitely read them from DB2 9.7 and older.