Skip to content

Data

Read or write any data source: a SQL query or a visual table builder.

What it is

Read or write ANY data source through one node: write SQL or build a query visually.

How it works

Pick a connection (the connection IS the engine - Postgres in the open edition). Then either write SQL (Validate checks it against the live database and publishes the result columns as downstream variables like {{step.data.row.email}}) or Build a query: pick a real table, an operation (select/insert/update/delete) and filters - no SQL needed.

Output

rows (capped ~1,000 rows / 256 KB so results fit through the workflow), rowCount, truncated, and row (the first record, for single-record references). Or set Output to reference to hand downstream a small dataset handle instead of rows - a later Data step opens it with "Open a dataset".

Gotchas

  • Use $1, $2 parameter placeholders in SQL; values never concatenate into the query.
  • Big results truncate at the cap - larger data belongs in a warehouse engine (Helix edition).

Configuration reference

  • connectorId - The data source connection to run against.
  • query - A SQL query, or build one visually. Its rows become this step's output. Use $1, $2 for parameter values.
  • output - Rows: the result set, inline (capped). Reference: a small handle a later step can open on demand - it never hits the payload limit and is how big data flows on Tachyon.
  • sourceRef - Materialise a dataset an earlier step handed on. Reference its dataset output, e.g. {{step.data.dataset}}.

Output fields

  • rows
  • rowCount
  • totalRows
  • truncated
  • dataset