Technology
PostgreSQL
Transactional integrity, JSON when you need it, full-text and vector search in the same engine — and an operational track record measured in decades.
Start here, and justify leaving
Postgres handles document storage, queuing, full-text search and vector similarity competently enough that a great many systems need exactly one datastore.
Every additional store adds a consistency boundary, a backup story and an on-call surface. We add one when the workload genuinely demands it, and we can usually show the query that proves it.
Practice
How we use it
Schema design
Constraints in the database, because application-layer validation is a suggestion that a second writer will ignore.
Query performance
Index strategy and plan analysis, with slow queries surfaced in monitoring rather than in a support ticket.
Safe migrations
Zero-downtime schema changes with expand-and-contract, tested against production-scale data volumes.
Beyond relational
JSONB, full-text and pgvector used where they remove a service rather than where they look clever.
Judgement
When PostgreSQL is the right call
And when it is not. A technology page that only lists strengths is a brochure.
Reach for it when
- Effectively any transactional workload
- Mixed relational, document and search requirements in one system
- Vector search at moderate scale alongside the operational data
Look elsewhere when
- Very high-volume time-series or analytical scans, where a columnar store wins
- Write throughput beyond what a single primary can absorb after tuning
Sectors
Where we run it
Also relevant