All articles

We’re starting with databases and storage. These are the working drafts you can read now; planned outlines are in the relational collection.

Start here

Databases and storage: a field guide

The main approaches to storing data, the problems they solve, and when to choose them.
Relational databases: an introduction

Tables, relationships, queries and reliable changes.

Start the SQL practice

Open PostgreSQL, load the shop records and choose the setup for each article.

Designing a relational schema

Work from an application's facts to tables, relationships, and a design that survives change.
Constraints

Make the database enforce your schema's rules, and recognise where those rules stop.
Tables and JSON

Choose what belongs in columns, related rows, and a document inside a record.

Queries and joins

Follow records through filters, joins and totals, and see where extra result rows come from.
Pagination and large results

Compare ways to continue through a result when records can change between pages.

What belongs in one transaction?

Choose which changes need to succeed or fail together.
When two requests change the same data

Explain how overlapping requests can break an application rule.
Isolation and snapshots

See which changes a transaction can read, and why a stable view does not prevent every race.
Handling waits, deadlocks, and retries

Respond correctly when a transaction cannot proceed or its outcome is unknown.

How a relational database keeps an order together

A worked checkout, from related records to concurrent purchases and recovery.

How a column store works

Column layouts, sorting, writes and replication, with four experiments.
How document boundaries shape a database

Nested records, shared details and the consequences of choosing a document boundary.
Choosing keys for partitioned data

Partition keys, sort order and the cost of another access pattern.
Keeping a transaction together across machines

Partitions, replicas and coordination in a distributed SQL database.
Keeping a copy of the answer

Cache hits, invalidation, expiry and eviction.
Finding the blue mug

Search indexes, tokenisation, ranking and vector filtering.
How a graph database follows relationships

Traversing connections, handling cycles and updating a graph.
What object storage gives you

Object keys, photographs and publishing a complete export.

Can Postgres handle vector search for my application?

A dated assessment of pgvector, retrieval quality and operating costs.