Hi - I'm trying to get my head around DMS, and how on-going replication works with a MSSQL RDS source.
As I understand it, on-going replication uses the fn_dblog function to read from the transaction log, so DMS can sync changes to the target. It opens a transaction against the awsdms_truncation_safeguard table to stop the transaction log being prematurely truncated.
Why then is it necessary to enable CDC on the SQL Server? Unless I've missed something obvious, DMS is not consuming the change tables which SQL Server maintains when CDC is enabled? Why is the CDC pollingInterval important for DMS? I ask because I have DMS running against a highly transactional table, and I'm having to manipulate pollingInterval, maxtrans, and maxscans to stop my transaction log filling up, and my lack of understanding of the process may be leading me to set them incorrectly. As I said, I don't even see why CDC is necessary....
