If your data already lives in a Postgres database or behind an Iceberg REST catalog (a common way for data lakes to list their tables), you don’t have to export it to ask Crunch about it. You link the tables, and Crunch reads them where they are — nothing is copied. That keeps one copy of your data, in the place your team already looks after.
What you’ll need
- A Team plan (see pricing).
- Sign-in details for your Postgres database or Iceberg REST catalog. Use a read-only user, so Crunch can read your tables but never change them.
Add a connection
- Open the Dataset Explorer (the database icon in the chat header) and find Connections.
- Click + Add connection, give it a name and choose the kind:
- Postgres — host, port, database, user, password, and optionally the schemas it may read.
- Iceberg REST — catalog URI, warehouse, and how it signs in (OAuth client, bearer token or none). Optionally add the namespaces it may read, and a storage key if your catalog doesn’t vend storage credentials (hand out the keys to the files itself).
- Click Add connection.
Link a table
- Click Link dataset, choose the connection and enter the table:
schema.table, ornamespace.tablefor Iceberg. - Click Read columns.
- Name the dataset and tick the columns that make a row unique. If you tick none, every row is treated as distinct. Set the time and user columns too, if the table has them.
- Click Link dataset.
How to tell it worked
The table appears in the datasets picker as a linked source, ready to pick in any conversation.
Good to know
- A linked table can’t be your clickstream, and it has no snapshots: Crunch reads it as it is now.
- Heads up: Removing a connection removes its linked datasets too. Conversations that used them can no longer read them.
Related
- Datasets — pick a linked table for a conversation.
- On-prem sync connector — for a database behind your firewall.
- Data home and lineage — see where every dataset comes from.