DB

Databases and persistence: the systems that store and serve a product's data.

This hub is the database layer of the directory: relational engines, document stores, vector databases, embedded files and serverless platforms, plus the clients and tooling that sit between an application and its data.

Why it matters to an indie builder: the database is the decision with the longest shadow. Data model, consistency and migration path chosen early shape everything downstream, and the modern field rewards a simple heuristic, start relational unless the domain says otherwise, and let access patterns, not fashion, trigger the specialized stores. The interesting shift is operational, serverless and branchable databases have collapsed the price of doing it right from day one.

A realistic pattern: Postgres as the primary with migrations in the repository, Redis beside it for the hot paths, pgvector or a dedicated engine when semantic search arrives, and SQLite embedded where the product lives on a device. Backups are scheduled and restore-tested, which is the difference between a policy and a hope.

The trade-offs: every specialized store adds a failure mode and a skill, and polyglot persistence is a maturity problem in disguise. The blocks here compare engines on the axes that matter, operational weight, scaling shape and exit cost.