Enterno.io scanned open port signatures + API response fingerprints to identify the primary database for top-50k web apps. PostgreSQL — 42% (leader, +5% YoY). MySQL — 31% (-3% YoY, slow decline). MongoDB — 15% (stable). SQLite — 7% (rising for edge). Redis as-primary — 2%. CockroachDB/YugabyteDB (distributed SQL) — 1% but growing.
Below: key findings, platform breakdown, implications, methodology, FAQ.
| Metric | Pass/Value | Median | p75 |
|---|---|---|---|
| PostgreSQL | 42% | — | — |
| MySQL / MariaDB | 31% | — | — |
| MongoDB | 15% | — | — |
| SQLite (edge + small) | 7% | — | — |
| Redis as-primary | 2% | — | — |
| Microsoft SQL Server | 1.8% | — | — |
| Oracle | 0.6% | — | — |
| CockroachDB / YugabyteDB | 1% | — | — |
| Platform | Share | Detail | — |
|---|---|---|---|
| PostgreSQL (self-hosted) | 26% | YoY: +2% | — |
| AWS RDS Postgres | 9% | YoY: +2% | — |
| Supabase (managed Postgres) | 4% | YoY: +1% | — |
| Neon (serverless Postgres) | 3% | YoY: +2% (new) | — |
| MySQL self-hosted | 22% | YoY: -2% | — |
| MongoDB Atlas | 9% | YoY: flat | — |
| Turso / LibSQL (edge) | 2% | YoY: +1.5% (edge) | — |
Top-50k sites with a detectable backend (not static). Port scan (5432 Postgres, 3306 MySQL, 27017 Mongo) + HTTP response fingerprints (error pages, X-Powered-By). Cloud DB detection via CNAME patterns (rds.amazonaws.com, supabase.co, neon.tech). YoY vs March 2025.
Postgres — default for new project (better features, JSON, partial indexes, better defaults). MySQL — if legacy team expertise or simple CRUD.
For multi-tenant apps with fluid schemas (chat, IoT, analytics). For relational data — Postgres with JSONB is superior.
Yes. Neon, Supabase, Vercel Postgres used in production. Main risks — cold start + pricing surprise on spikes.
Impossible 100% externally. Indirectly: error messages, response time patterns, admin panel paths. Enterno — API response timing + type of 500 errors.