Module 7.1: Inside the Cloud Warehouse
The MPP warehouse from the inside: nodes and slices and why a query finishes at the slowest slice, distribution and sort keys read off simulated table stats and query plans (including the DS_DIST_BOTH step interviewers ask you to spot), and the loading and billing models that decide what a query costs.
Inside an MPP Warehouse: Nodes, Slices, and the Slowest Slice
A leader node plans, compute nodes execute, and every distributed table is sharded across every slice. The query runs on all of them at once and finishes at the speed of the slowest one, which is why skew in slice-level row counts is a performance bug.
Distribution Styles, Sort Keys, and Reading the Query Plan
KEY co-locates joins, ALL copies small dimensions everywhere, EVEN round-robins facts nobody joins on. The plan tells you when you chose wrong, and DS_DIST_BOTH is the step interviewers ask you to spot.
COPY from Stage, Never Row Inserts, and the Two Billing Models
Warehouses ingest by pointing COPY at a stage of files in object storage and loading them in parallel across slices. Then the money: per-TB-scanned and per-second-compute price the same workload very differently.
Module 7.2: Dimensional Modeling Rehearsal and the Missing Debate
Review-plus-extension, not a reteach. SQL Levels 3 to 5 already taught grain, additivity, and SCD Type 2; this module compresses each into one review screen and grades what interviews actually separate on: an executable grain audit, the wrong-sum beside the right-sum, a timed SCD2 change-batch applier, and the Kimball versus One Big Table decision with its 10 GB join-side heuristic.
Grain Rehearsal: Say the Sentence, Then Prove It in SQL
Review declared grain in one screen, then do the new work: audit a raw ride-share extract against its declared grain with executable SQL before any star gets built.
Semi-Additive Rehearsal: The Wrong Sum and the Right Sum, Side by Side
Review measure additivity in one screen, then grade the new artifact: a single query that shows the naive cross-day SUM next to the correct as-of total so the delta is visible.
SCD2 Under the Clock: Apply the Change Batch
A timed rehearsal of the interview's stock modeling scenario: apply an incoming multi-driver change batch to a Type 2 dimension with a script that is safe to run twice.
Kimball vs One Big Table: The 10 GB Join-Side Heuristic
Decide star schema versus One Big Table per data mart using consumer needs and the ~10 GB join-side heuristic, then price what the denormalization actually costs.
Module 7.3: The Lakehouse and Open Table Formats
What Iceberg adds on top of the files-plus-catalog world Level 6 built: transactions and a metadata tree of snapshots and manifests, MERGE semantics that survive a rerun, time travel straight out of the metadata, and the compaction plan that fixes the small files streaming writes leave behind.
What Iceberg Solves: Snapshots, Manifests, and the Metadata Tree
What an open table format adds on top of raw Parquet in object storage, and how to read Iceberg's snapshot and file metadata to see exactly which files a query would scan.
MERGE Semantics: Upserts That Survive a Rerun
Implement warehouse MERGE semantics with INSERT ... ON CONFLICT DO UPDATE, guard the update so a late-arriving old row cannot overwrite a newer one, collapse a redelivered batch to one row per key first, and prove the load is idempotent by running it twice.
Time Travel, Snapshot Expiry, and the Compaction Plan
Reconstruct a table as of an earlier snapshot from metadata alone, then pick the compaction candidates that fix the small-files problem streaming writes create.
Module 7.4: Platform Choice and the Microsoft Stack
The judgment layer: translate the whole platform into Azure and Fabric vocabulary for the enterprise segment that hires on it, decide warehouse versus lakehouse per workload instead of by dogma, and close the level by pricing a month of work under three billing archetypes with no scaffold.
The Same Platform, Twice: AWS to Azure and Fabric
Translate every layer of the data platform between AWS, Azure, and Fabric names, and read where Azure demand actually concentrates.
Warehouse, Lakehouse, or Both: A Decision Framework
Choose warehouse, lakehouse, or a hybrid tier for a workload using a stated framework instead of dogma.
Snowflake, BigQuery, Fabric: The 2026 Landscape and the Cost Showdown
Speak the 2026 warehouse market in one screen, then price a month of workloads under three billing archetypes and pick the cheapest platform per workload.