The following diagram shows the basic architecture of Qlik Replicate.
Qlik Replicate System Architecture
In the initial load process, Qlik Replicate reads a filtered stream of rows (with relevant columns only) and passes them to the transformation process for further filtering and subsequent writing to the target endpoint (in the expected output format).
The CDC process obtains a stream of filtered events or changes in data or metadata from the transaction log file. It then buffers all changes for a given transaction into a single unit before forwarding them to the target when the transaction commits. During the initial load process, CDC also buffers all changes that occur within a transaction until all affected tables have been loaded.
The Designer/Console server, which is part of the Replication server, is a Web-based application that serves as the user interface for dealing with designing or modifying the replication system and displaying and controlling its operation.