SQL Workbench

Data Engineering → Workbench → Query Editor is the SQL IDE. It has two engines:

Tab	Engine	Runs as	Governance
Lakehouse	Trino	You	✅ Your row filters, column masks and access policies all apply
DN Warehouse	StarRocks	A shared workspace role	Workspace isolation only. Not per-user.

For sensitive data, use the Lakehouse tab. It is the fully per-user governed SQL surface.

Catalogs you'll see (Lakehouse tab)

Catalog	Contents
lakehouse (also iceberg)	All Iceberg tables: bronze, silver, gold, sandboxes
dnw	Your DN Warehouse database
postgres	Workspace operational Postgres. Internal; read-only use.
system, tpch, tpcds	Built-in and benchmark data

In the DN Warehouse, your database is <safe> (the workspace slug with underscores). Lakehouse tables are also visible there through the external catalog <safe>_iceberg, e.g. SELECT … FROM <safe>_iceberg.gold.sales_daily.

Running queries

• Ctrl/⌘ + Enter runs the query.
• Pick a catalog and schema, and bare table names in FROM / JOIN are qualified for you. Fully qualifying (lakehouse.gold.sales) always works.
• Limit dropdown: 50 / 100 / 1,000 / 10,000. Interactive results are capped at 10,000 rows. The grid pages 500 rows at a time.
• Copy copies the returned result as CSV. The Voir side panel can Explain this query or Optimize for performance.
• On the DN Warehouse tab you may first need Connect to Warehouse.

Export CSV (full, governed exports)

Export CSV re-runs your last Lakehouse query on the server as you and streams every row to a file. It is not limited to the 10k rows in the grid.

Rule	Value
Engines	Lakehouse (Trino) only. Disabled on DN Warehouse ("Run a Lakehouse query first").
Statements	SELECT / WITH only
Governance	Same row filters and masks as the grid
Size / time	1 GiB / 20 min per export
Rate	5 per minute, 50 per hour
Your LIMIT	Respected
Failure mid-stream	Error, no file. An engine or network failure never leaves a partial file behind.
⚠️ Hitting the 1 GiB cap	The file stops at 1 GiB with no warning. It looks complete but isn't. If an export is close to 1 GiB, assume it was cut off: add filters, split by date range, or write a table instead.

For bigger extracts, write a table (dbt or notebook) and ask about bulk-export options.

Connecting DBeaver, JDBC and BI tools

External SQL clients connect to the query engine with your platform username and password. Queries run as you, so governance applies.

Setting	Value
Driver	Trino
Host	query.<domain> (your admin will confirm the exact host)
Port	443, SSL on
JDBC URL	jdbc:trino://query.<domain>:443/lakehouse?SSL=true
User / password	Your platform credentials
Catalog	lakehouse

• Complete any pending account actions in the browser first. Clients can't log in while one is pending.
• ⚠️ Update saved passwords straight after changing yours. Repeated failures lock your account.

Writing fast queries

• Filter on time. Always add a date predicate on large tables.
• Select columns, not *. Iceberg is columnar, so fewer columns means less data scanned.
• Joins: put the large table first (left) and the small or driving set second, and filter both sides before joining. Very large joins can hit the per-query memory limit (Query exceeded per-node memory limit).
• Run ANALYZE lakehouse.<schema>.<table> on large tables you join often. Table statistics help the planner.
• Prefer gold. If you repeat the same heavy query, make it a dbt gold model or a cube.
• The first query on a freshly created catalog can time out while metadata warms up. Retry.

Checking a table's partitions

SELECT * FROM lakehouse.bronze_crm."orders$partitions";
SELECT * FROM lakehouse.bronze_crm."orders$files" LIMIT 20;   -- file sizes

If $partitions returns 0 rows, or a single partition with a NULL key while the table has data, the table is affected by the partition case issue (Limits & Known Issues). Results are still correct, but filter on real columns, not partitions.

Troubleshooting

Symptom	Cause	Do
Only 10,000 rows	Interactive cap	Export CSV
Export CSV disabled	On DN Warehouse tab, or no Lakehouse query yet	Switch to Lakehouse and run first
Export rate limit reached	5/min, 50/h	Wait
Export 403 with explanation	Your identity couldn't be established for the export	Contact your admin
Access Denied: Cannot select from …	No policy grants you that table	Ask a workspace admin (Access & Policies)
Schema browser shows only default or nothing	An over-broad policy is hiding schemas from you	Contact your admin
DN Warehouse shows no databases	Warehouse not connected, or grants incomplete	Connect to Warehouse; if it persists, contact your admin
Access Denied on mv_* tables	Cube MVs are only readable through the semantic layer	Use the cube (Semantic Layer (Cubes))
TABLE_NOT_FOUND on a CTE name	Older release auto-qualified CTE names	Fully qualify real tables; ask for an upgrade
Exceeded per-node memory limit	Join or aggregation too big	Filter earlier, reorder the join, pre-aggregate in dbt