Cursor Column Support for Reverse ETL Syncs Beta

Previously, RudderStack’s Reverse ETL Upsert mode was diff-based, meaning it compared every column and row to detect changes from your source table in order to determine what records needed to synced. This led to longer sync times and higher warehouse costs, particularly for large tables.

With Cursor Column support, you can leverage a timestamp value checkpoint for the diff. This optimizes Reverse ETL syncs by allowing you to specify a comparable timestamp column (for example, updated_at) to track incremental changes in Upsert mode. RudderStack will then query only records with a cursor value greater than the last checkpoint, avoiding resource-intensive, full-table comparisons.

By reducing the computational load of warehouse queries, this feature lowers operational costs and speeds up syncs, especially for large datasets.

Setup guide

To enable Cursor Column support in Upsert mode:

1. Create a new destination connected to a warehouse table or SQL model as a source.
2. After adding the destination credentials and mappings, choose Upsert as your sync mode.
3. Turn on the Cursor Column toggle.
4. Choose the timestamp column you want to use as the cursor column.
5. Initiate your first sync.

RudderStack will perform a full sync on the first run.

After the first run, RudderStack will use the cursor column to identify any new or updated records that arrive in the source table and sync those records on any subsequent sync jobs.

See our Cursor Column Support docs for more information on this feature.

Screenshots

Questions? Contact us by email or on Slack