Run a prune
Run prune opens straight after you create a data set. To come back to it later, click Prune on the data set's card.
Settings
| Setting | What it means |
|---|---|
| Sample of starting rows | How much of the starting tables to keep, from 1% to 100%. Related rows follow automatically. |
| Follow relationships | On: bring related rows from linked tables too. Off: starting tables only. |
| Row filter | Optional. One condition per starting table, e.g. Text starts with…, Year is…, Number is greater than…, Is empty (NULL). The SQL preview shows exactly what will run. |
| More options → Depth limit | How many links to follow from each starting table. 0 follows every link. |
| More options → Table files | .csv or .json files per table. |
A summary line, such as Online shop · 6 tables, shows what will run. Click Run prune. Progress shows on the page, and the result opens when it's done.
DataFactory remembers your settings for each data set. Run again on a result starts from that run's settings.
Every relationship stays intact
If a kept order references a customer, that customer is kept too, even if it falls outside your sample. That's why the achieved size can differ a little from the target.
Good to know
- One prune per data set runs at a time (A prune is already running for this data set.). Each source database allows two at once.
- Very large tables use database statistics to estimate counts, so progress figures may be approximate.
- Prune only reads from the source. Nothing is changed or deleted there.