Every Writeback table loads its rows with a sync type: a full load reads the whole source, and an incremental load reads only recent rows. This page covers both and where the choice lives.
The Publish Window sits on General, and it appears only on tables of RabbitMQ destinations. Sync mode offers two loads:
| Sync mode | What each run reads |
|---|---|
| Full snapshot | Every row the source returns. This is the default mode. |
| Incremental | Only the rows inside the publish window, up to the date and time of the run. |
With Incremental, two more fields appear: Increment column, which chooses the column the window follows, and Minus days, which sets how far back the window reaches. Each change saves at once. See Publish to RabbitMQ for the window and its fields in full.
Writeback itself offers this one choice: a full or an incremental load. The wider set of load patterns in DataMind OS, the incremental reloads and their loop, the upsert types, the snapshot types and one-time loads, belongs to the tables' own sync settings as data loads into the Lake. For the patterns and how to pick one for a table, see Choose a sync type. The tables a Writeback table reads keep their sync settings where they live.
Yes. Until you change the load, every run reads the whole source. A table on a RabbitMQ destination can switch to an incremental load; a table on any other destination type stays on the full load.
A full load reads every row the source returns on every run. An incremental load reads only the rows inside the publish window, so each run covers recent rows instead of the whole table. See Publish to RabbitMQ.
It appears only on tables of RabbitMQ destinations. For a table on a database or Google BigQuery destination, each run reads everything its source returns. To send fewer rows, narrow the source instead: turn Table Selection off on General and write a query that selects only the rows you need. See Configure a Writeback table.
No. The load decides which rows a run reads. The columns and values written come from the mapping on Columns and the Data Type of each mapping row. See Map columns.
Yes. On a RabbitMQ destination, set Sync mode back to Full snapshot, and every run reads all the rows again. A table on any other destination type is already on a full load.
No. A change applies to later runs, and the rows already in the destination stay as they are. See Writeback for what each run adds.
On that table itself, where it lives. For a Lake table, Sync Type sits under Synchronization on the table's General step. See Choose a sync type. Your next Writeback runs read the new data.
The Increment column list offers the table's columns, so any of them can be picked. For the window to follow your changes, choose a column that changes when a row changes, such as an update timestamp. See Publish to RabbitMQ.
No. Sync mode in the Publish Window sets whether each run reads all rows or a recent window. Sync Schema, on a destination's right-click menu, checks the destination for new columns on the tables already in Writeback and adds them. See Add and manage destinations.