Skip to main content
ExplanationApplies to: Unified engine deployments
Build something small, inspect its result, then change one thing. These paths take you from a first query to a scheduled workload.

​
Learning paths

  • Run your first analysis: Query six orders, check the totals, and save your analysis. Requires: SQL execution.
  • Query your first dataset: Read, shape, and sort a small dataset defined inside the query. Requires: SQL execution only.
  • Filter and classify data: Keep the rows you need and label them with clear rules. Requires: SQL execution only.
  • Handle missing values: Understand NULL, supply defaults, and count missing values correctly. Requires: SQL execution only.
  • Summarize data: Calculate totals and compare groups without creating a table. Requires: SQL execution only.
  • Combine datasets: Match customers and orders, and find records with no match. Requires: SQL execution only.
  • Build queries in steps: Use named steps and subqueries to make an analysis easier to read. Requires: SQL execution only.
  • Compare lists: Combine, deduplicate, and compare two small lists. Requires: SQL execution only.
  • Clean text and convert types: Normalize text, extract fields, and convert values safely. Requires: SQL execution only.
  • Work with dates and time: Use fixed dates so every run has a clear expected result. Requires: SQL execution only.
  • Rank and compare rows: Use window functions to keep detail rows while comparing related values. Requires: SQL execution only.
  • Read JSON and nested values: Extract useful scalar values from JSON, arrays, and structs. Requires: SQL execution only.
  • Solve complete analysis problems: Combine the earlier lessons to answer practical questions with small inputs. Requires: SQL execution only.
  • Understand MERGE matches: Match existing and incoming customers, then update an existing customer and insert a new one. Requires: Writable Iceberg catalog with MERGE support.
  • Choose which rows MERGE updates: Update selected fields and apply conditions without changing the meaning of a match. Requires: Writable Iceberg catalog with MERGE support.
  • Combine MERGE actions: Use update, insert, and delete clauses together and observe clause order. Requires: Writable Iceberg catalog with MERGE support.
  • Synchronise a complete source snapshot: Mark or remove target customers that are absent from a complete source snapshot. Requires: Writable Iceberg catalog with MERGE support.
  • Prepare a reliable MERGE source: Choose one source row per key, match composite keys, and handle NULL keys explicitly. Requires: Writable Iceberg catalog with MERGE support.
  • Make repeated MERGE batches predictable: Compare a version-guarded merge with an additive update when a batch is repeated. Requires: Writable Iceberg catalog with MERGE support.
  • Apply customer change events: Compare physical deletion with a stored deletion marker and reject an older update. Requires: Writable Iceberg catalog with MERGE support.
  • Run MERGE incrementally: Apply appended change batches with saved source progress and verify a no-input run. Requires: Incremental SQL workloads and a writable Iceberg catalog.
  • Secrets: Create local secrets and external references with SQL, then inspect and rotate them. Requires: Secrets service; configured providers for external values.
  • Credentials: Choose an authentication method, define it with SQL, and verify its consumer. Requires: Credentials service and the identity setup for your chosen method.
  • Connect your data: Connect a catalog, inspect a table, and verify data access. Requires: An existing catalog service.
  • Build a dashboard: Turn the sample orders into a chart with a region filter. Requires: Dashboards and SQL execution.
  • Build a notebook: Combine SQL and explanation, then compare parameter values. Requires: SQL notebooks.
  • Automate an analysis: Connect two tasks, publish a summary, and schedule it. Requires: SQL workloads and a writable Iceberg catalog.
  • Process new data: Append source rows and verify that saved progress advances. Requires: Incremental SQL and a writable Iceberg catalog.

​
Choose your starting point

Start with Run your first analysis. It uses six orders included in the SQL. You do not need a catalog or an external dataset. The dashboard and notebook paths use the same orders. Then follow the 12 SQL practice lessons, from Query your first dataset to Solve complete analysis problems. They contain 76 complete examples with inline data, expected result tables, and small exercises. Start any example on its own. Read Secrets to store values or reference an external secret provider. Then read Credentials to choose an authentication method and define it with SQL. Both pages include examples for every supported type. Choose Connect your data when you have a catalog service to connect. The automation and incremental paths need a writable Iceberg catalog. Each one creates separate tables so you can follow it on its own.

​
Learn MERGE

Follow the MERGE path to update, insert, and delete rows; apply conditions and clause order; synchronise a complete snapshot; prepare source keys; and compare repeat runs. Continue with change events and incremental workloads. These eight lessons need a writable Iceberg catalog. They use small inline customer datasets and exact expected rows.

​
How to use a lesson

  1. Read Before you start and check the required capabilities.
  2. Follow the steps in order in your product workspace or client. Check the result at each stage.
  3. Complete Change one thing and compare the result.
  4. Follow Clean up when you finish.
Run related statements in order in the same session. Each statement has its own result. A CTE belongs to one statement, so repeat its WITH block when another statement needs the same input. Review the selected catalog, namespace, and compute before running. SQL that creates or changes tables performs real work on that selected catalog.
Last modified on November 4, 2026