6.1. Index¶

Each scenario ships paired Zeppelin and Jupyter notebooks with equivalent intent. Seventeen Spark scenarios pair Scala with PySpark; the two Trino scenarios pair %trino SQL with the Python Trino client. The manifest-owned walkthroughs below are canonical. Spark pages show side-by-side language parity, while Trino pages document query/result equivalence without calling SQL “Scala” or the Python client “PySpark.”

The execution-mode matrix records which notebook scenarios also have production orchestration. A paired notebook is always an executable teaching surface; it does not imply that a production DAG exists. For the two Trino pairs, notebook CTAS cells are an educational direct-write path and do not enforce production provenance, snapshot checks, or serialization. Use tpch_bi_query or nyc_taxi_trino_daily for production; their durable metadata-DB XCom result is not an Iceberg table. The TPC-H star-schema, MovieLens feature-engineering, and GH Archive json_flatten-gh_archive-spark-iceberg / sessionization-gh_archive-spark-iceberg walkthroughs document notebook-equivalent transforms behind the production tpch_star_schema, movielens_feature_pipeline, and gh_archive_flatten_sessionization applications. Their notebooks are educational parity surfaces, not supported production write paths, because they can directly replace the same tables without production provenance, validation, or serialization.

Update a walkthrough alongside its source notebooks, then run make docs-check. The aggregate gate verifies that every paired scenario has exactly one manifest-owned walkthrough and projects it to the site and wiki.

6.1. Index¶

1. Batch¶

2. Streaming¶

3. Quality / Modeling¶

4. Ops¶

5. SQL / Analytics¶

6. Semi-structured¶

7. See Also¶