Module 06
Data modeling & storage
- SQL vs key-value / document
List your queries and required guarantees and watch each candidate data model light up green or red for each one.
- Normalization vs denormalization & projections
Update a denormalized field and watch the fan-out of writes needed to keep every read-optimized copy correct.
- Indexing (B-tree, hash)
Run a query with and without an index and watch the scan path through a B-tree next to a full table scan, with page-read counts.
- Sharding, partitioning & hot partitions
Shard a 500M-row users table by user_id or by region and watch load heatmaps, cross-shard queries, and a hot partition appear.
- Time-series, graph & columnar stores
Run fraud-detection queries (time windows, shared-device traversals, analytic scans) against each store type and compare the work each one does.
- Object vs block storage
Place photo originals, thumbnails, and metadata onto S3-like or EBS-like storage and see latency, cost, and access-pattern fit.
- Inverted indexes & filtered search
Type a query, watch term posting lists intersect, then add location and salary filters and see where structured filtering fits in.