Oracle asynchronous change data capture

Oracle Database Tips by Donald Burleson

The first thing we want to do then is set up the source and target databases to work with change data capture. The source database has to be in ARCHIVELOG mode, and both databases have to have GLOBAL_NAMES=TRUE set, and in particular the source database has to have the following parameters set:

• compatible='9.2.0.0.0'   or higher
• global_names=TRUE
• job_queue_processes=2
• log_archive_dest_1='LOCATION=<full path>'
• log_archive_format='%t_%s.dbf'
• log_parallelism=1
• open_links=4
• open_links_per_instance=4
• parallel_max_servers=3
• processes=150
• undo_retention=3600

(note that I couldn't set LOG_PARALLELISM=1 on my 10gR1 database as this only existings for 9i, so I left this as it was, and I also couldn't start up the database afterwards when I set the LOG_ARCHIVE_FORMAT to '%t_%s.dbf', it complained about wanting a '%r_' in there as well, so I left that one as the default value.)

The destination database needs to have the following parameters set:

• compatible='10.2.0.1.0'
• global_names=TRUE
• open_links=4
• open_links_per_instance=4
• parallel_max_servers=2
• processes=150
• java_pool_size=50M
• streams_pool_size=50M