🧭 NoSQL Concepts
What the label actually covers — the five data models and which question each answers cheaply, query-first modelling and what replaces the join, partition keys, quorums, and CAP taught correctly. Then the honest closing argument: one PostgreSQL does four of the five, and the fifth database is somebody's pager.
No lessons published yet — this is the roadmap, so you can see exactly what the course will cover before it lands.
Before you start: Data Modeling — Series 1. You cannot appreciate query-first modelling without first knowing the discipline it reverses. Keys, relationships and normalization are what Series 1 · 10 here deliberately turns upside down, and the reversal only teaches anything to someone who knows the original rule and why it is right.
Short on time? The 19 episodes marked core are the shortest path to being productive at work. Start there and pick up the rest whenever a job asks for them.
Series 1 · Why Not Just Relational?
the pressures that break the default, and the modelling reversal
- 01
Where the Relational Default Stops — Four Pressures, and One That Is Not
coreintermediatecoming soon - 02
What NoSQL Means — a Label for a Family That Shares Almost Nothing
coreintermediatecoming soon - 03
Query-First Modelling — Design for the Question, Not for the Entity
coreintermediatecoming soon - 04
What Replaces the Join — Embed, Reference, Duplicate or Precompute
coreintermediatecoming soon - 05
Project — Model the Same Business Relationally and Then for Its Queries
coreintermediate#projectcoming soon
Series 2 · The Five Shapes
what each model is for, and which question it answers cheaply
- 01
Key-Value — the Fastest Thing There Is, If You Already Know the Key
coreintermediatecoming soon - 02
Document — an Aggregate Stored and Read as One Thing
coreintermediatecoming soon - 03
Wide-Column — a Partition Key, and Everything Under It in Order
coreintermediatecoming soon - 04
Search — the Inverted Index, and Why Relevance Is Not a Filter
coreintermediatecoming soon - 05
Graph — When the Relationship Is the Data, and Depth Is the Question
coreintermediatecoming soon - 06
Choosing a Shape — Starting From the Questions, Not From the Products
coreintermediatecoming soon
Series 3 · Distributed, and What It Costs
partition keys, quorums, and consistency taught honestly
- 01
Partition Keys — the One Decision Everything Downstream Inherits
coreadvancedcoming soon - 02
Replication and Quorums — Why R + W > N Is the Whole Argument
coreadvanced#decandia-dynamocoming soon - 03
CAP, Correctly — and Why PACELC Is the Question You Actually Live With
coreadvancedcoming soon - 04
Eventual Consistency at a Desk — the Support Ticket, and What It Is Called
coreadvancedcoming soon - 05
Transactions — One Document at a Time, and What Arrived Since
advancedcoming soon - 06
Secondary Indexes — Local, Global, and Why the Advice Is Another Table
advancedcoming soon - 07
Project — Predict What a Cluster Returns During a Partition, Then Check
coreadvanced#projectcoming soon
Series 4 · Choosing, and Living With It
the relational case put properly, and the cost of the fifth database
- 01
One Postgres Does More Than You Think — JSONB, Full-Text, Geo and Vectors
coreadvancedcoming soon - 02
Distributed SQL — the Category That Took the Best Argument Back
advancedcoming soon - 03
Polyglot Persistence — the Fifth Database, and Who Is On Call For It
coreadvancedcoming soon - 04
Migrating On and Off — the Model Does Not Survive the Move
advancedcoming soon - 05
Project — Write the Memo That Chooses a Database, With Its Trade-offs Named
coreadvanced#projectcoming soon
Where this leads
- Apache Cassandra — starts from Series 1, Series 2, and Series 3.
- Elasticsearch — starts from Series 1, and Series 2.
- MongoDB — starts from Series 1, and Series 2.
- Neo4j — starts from Series 1, and Series 2.
- Redis — starts from Series 1, and Series 2.
Not here, on purpose
- Any one product, in depth — covered in MongoDB, Redis, Elasticsearch, Apache Cassandra and Neo4j, each of which assumes this course. Here every product is an example of a shape, never the subject
- Relational modelling, normalization and the star schema — covered in Learn Data Modeling, which this course assumes and argues with rather than repeating
- Vector databases and similarity search — covered in AI for Data, Series 2, which teaches embeddings and argues for vector storage inside the database you already run. In this course, Series 2 · 25 names the shape and points there
- Distributed SQL — CockroachDB, YugabyteDB, TiDB and Spanner — covered in Distributed SQL, which is a relational course and is filed with the relational engines. In this course, Series 4 · 05 names the category and states why it matters to this decision, and stops there
- Cloud-vendor proprietary databases — DynamoDB and its equivalents — not covered, deliberately. This catalogue teaches engines you can run yourself, and a managed service that exists on one cloud is a different kind of commitment. In this course, Series 2 · 00 and Series 2 · 10 name DynamoDB as an example of a shape, which is as far as it goes
- Running any of these in production — covered in Data Ops for the discipline. Each engine's own track owns its operational specifics
- Object storage as a data store — covered in Data Ops, Series 1 · 47. It is not a database and this course does not pretend it is one