When we reach for it
Almost every system we build. Constraints in the database are the last defence against bad data, and they hold when the application layer is bypassed by a migration, a script or a future service nobody has written yet.
PostgreSQL engineering
A relational database with transactional guarantees, JSON when you need it, full-text search and a query planner that will tell you what it is doing — our default, and the one we have to be argued out of.
What we reach for first
When we would argue against it
It is the wrong tool for genuinely enormous analytical scans, where a columnar warehouse is a different class of machine, and for a cache, where Redis exists. Neither is a reason to start elsewhere.
What it looks like in delivery
Migrations in the repository beside the code that needs them, foreign keys and check constraints written rather than assumed, and query plans read before an index is added. Most database performance work we are called into is a missing index and a query somebody let an ORM write.
Where this appears on the site
Published work that names it
- Rebuilding subscription billing and entitlements at SalteaCase study · architecture
- Removing hand-keyed identifiers from a retinal screening imaging pipelineCase study · architecture
- Closing the daily reconciliation break queue at ZafarinCase study · architecture
- Making offline receipt capture reliable in Vertigem's expense platformCase study · architecture
- Web developmentService · stack
- Custom softwareService · stack
- API developmentService · stack
Derived from the stacks published on those pages, not written here — so this list cannot claim something the page it points at does not say.
Working in PostgreSQL?
Tell us what it is running, what it costs you today, and what you need it to do next. A senior engineer will tell you what we would keep and what we would change.