Finally, after much procrastination, the dbreplicator project has released its first public version. This enables multi-master database replication between a variety of RDBMS's, and contains many performance enhancements over its predecessor. Key improvements include improved conflict handling, improved memory handling resulting in reduced memory requirements, faster sync and snapshots, cleanup of old shadow records, proper handling of columns with special characters, fixes to update trigger creation, a linux server mode (starting dbreplicator as a service) and debian packaging. For a complete listing of changes, see the changelogs.

Significant features include:

• Heterogeneous Replication - i.e., synchronizing the same database between multiple RDBMS's.
• Bi-Directional data synchronization - between any of the supported database backends, with subscriber-initiated synchronization
• Platform Independant - 100% java client and server components
• Application Independance - software need have no special knowledge of the replication, nor access its datasource any differently. This is great for applications to which you don't have source code access or rights to request changes
• Automatic Confilct Detection and Resolution - requires minimal involvement by database administrators. Basic conflict resolution is implemented as publisher wins/subscriber wins. Future versions will have more complex rule-based conflict resolution algorithms.
• Scheduling Facility - built into dbreplicator, the schedule can be specified in minutes, hours, etc., and runs the desired operation at the specified interval ad infinitum unless it encounters an error.
• Verbose Debugging - implemented using apache log4j, this can be quite flexible in terms of output methods and verbosity.
• Special Character Handling - DBReplicator handles characters with nonprinting ascii values by encoding the entire field in escaped unicode when creating sync xml files, and un-encoding at the opposite end. At present, those fields containing special characters must be specified in EncodeConfig.ini, but automatic handling is planned.
• Table Creation - if tables don't exist in subscriber databases, DBReplicator can automatically create them for you to match those in the publisher database.

Some things to note:

DBReplicator requires a Java Runtime Environment (JRE), and has only been tested on Sun's version. It has been tested with both Java5 (1.5.x) and Java6, and although it runs perfectly on both, it will only compile on Java5 due to some code which has been deprecated. Something to keep in mind if you're downloading the source or recompiling. Downloading is presently through sourceforge, the link to which is located here.

Brian