...
The Destination Database supports parallel commits to the same table;
The Source Database supports the reading by-chunks.
Important:It make sense to turn-on Chunking if the number of parallel processes more than one.
By default Chunking is turned-Off. To turn it On for specific table go to Workflow View → Config Tables and click on the chunking icon
...
Turn the chunking On for huge tables only.
Explanation: It’s not effectively to cut small tables on chunks and insert them concurrently in few threads because the concurrent insert (the switches between threads) takes time.
Don’t set the number of parallel processes more than 4 for databases what doesn’t supports the reading by-chunks.
Explanation: If your source databases MySQL and PostgreSQL what doesn’t allow to read only specific chunk (part of data) then each parallel process should get scan all records (to do “full table scan“), filter his specific chunk and send it. The spending time on the reading scanning all records for each parallel process make sense if the writing works a few times slow than reading onlythe scanning.