Replication server can perform some basic tranformations (via function strings) but as OLAP is a much more complicated design, you would want to use the ETL capability of the DMM to create the transformations to the OLAP model.
For now, you probably want 2 steps - 1. move the data from one PDM to another using DMM for replication only, 2. transform it using another DMM model for the ETL transformation steps from OLTP to OLAP.
You may want to use a second PDM as both the target for the replication and the source and target for the OLAP. You could even combine both extensions in a DMM and do it all in one DMM (that is up to you), but you would still want to separate replication functionality from ETL functionality.
One other thing to keep in mind - depending on the functionality of the target DBMS, you may not need the ETL capability for OLAP. Many DBMS engines (e.g. HANA, Oracle, etc) can create OLAP schemas/functionality directly from underlying tables, without needing external tranformations to persistent DBMS objects. In this way, you get immediate OLAP capability from the DBMS after only loading the base tables. Not all DBMS engines need star/snowflake schemas to create OLAP objects.
To take advantage of this, you may only need to replicate to a target DBMS with the same table schema and then use PD's multidimensional modeling capability to create OLAP cubes in the target PDM. Although you may need/want some transformations to a star/snowflake schema anyway - depending on your methodology - you might be able to accomplish this through ETL or just DBMS capabilities (e.g. stored procs/views).
PD's multidimensional modeling is built into the PDM capability.
Chris
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.