What I noticed
I ran into Horizontal Partitioning in Connect CDC, and it changed how I look at replication. It is not just table-to-table sync. You can route rows from the same source table to different targets based on data values.
How it works
Horizontal Partitioning uses Gate Conditions on target mappings. Each target has a condition, and CDC routes each row based on its values.
If a row changes and later matches a different condition, CDC can remove it from the old target and insert it into the new one. There is also an option to keep the old target row if needed.
Two things that mattered in practice:
- Gate Conditions are case-sensitive. If your data has mixed casing, use functions to normalize values before comparison.
- Scope tracking only works when the condition is based on columns in the source table. If you use system variables, CDC cannot detect previous values, so it will not remove rows from the old target.
Where it helps
I've seen this fit well in scenarios like:
- Splitting one table across regional targets
- Sending specific categories of rows to different systems
- Keeping target databases limited to only relevant data
- Building one-to-many distribution without extra logic outside CDC
Why it stands out
Most replication setups just keep tables in sync. This lets you apply business rules to decide where data goes, directly inside CDC processing. No extra pipelines or custom routing needed.
Have you used Horizontal Partitioning in your models, or are you handling routing outside CDC?
------------------------------
Adhitya Maya
*Precisely Software Inc.
------------------------------