Data Warehouse Patterns - SQL Server Integration Services Design Patterns - page 322

Figure 11-7 . CDC Source
The CDC Splitter breaks apart the data stream and sends rows out to the insert, up-
date, and delete outputs. From there, we'll write the update and delete rows to a staging
table so we can process those as high-performance, set-based operations. The insert re-
cords can go directly into the output table.
It's worth mentioning here that several CDC processing modes are available
through the CDC Source component. The example in Figure 11-7 illustrated the use of
the Net setting, which is the most common mode in most data warehouse scenarios.
However, depending on the ETL requirements and source system design, you may opt
for one of the other processing modes, as follows:
All: Lists each change in the source, not just the Net result
All with Old Values: Includes each change plus the old values for
updated records
Net with Update Mask: Is for monitoring changes to a specific
column in the monitored table
Net with Merge: Is similar to Net, but the output is optimized for
consumption by the T-SQL MERGE statement