Should I use PostgreSQL or MySQL?

PostgreSQL is our recommendation for any new project. It has superior support for JSON, full-text search, window functions, common table expressions and extensions (pgvector, PostGIS). MySQL has a larger legacy install base but PostgreSQL has overtaken it as the developer community's preferred relational database. The syntax is nearly identical — migration between them is straightforward. Unless you have a specific existing MySQL deployment, PostgreSQL is the correct choice.

How does pgvector work for AI applications?

pgvector is a PostgreSQL extension that adds a vector column type and similarity search operators (cosine similarity, L2 distance, inner product). You store embeddings (generated by OpenAI or Anthropic embedding models) alongside your regular relational data, then query by similarity. For most RAG applications with fewer than 5–10 million vectors, pgvector in your existing PostgreSQL database is sufficient — no separate vector database required.

Should I use Supabase or raw PostgreSQL?

Supabase adds auth, real-time subscriptions, file storage and an auto-generated API layer on top of PostgreSQL — significantly reducing backend code for standard CRUD applications. For projects that need these features, Supabase is faster to ship. For projects with complex server-side logic, existing infrastructure or specific deployment requirements, raw PostgreSQL with Prisma or Drizzle gives more control. Both run the same PostgreSQL underneath — you can migrate between them.

Can PostgreSQL scale to handle large applications?

Yes. PostgreSQL handles millions of rows efficiently with proper indexing and connection pooling (PgBouncer). Instagram, Shopify and many high-traffic platforms run PostgreSQL at scale. For read-heavy workloads, read replicas distribute load. For very large-scale write loads requiring horizontal sharding, Citus (now part of Azure) extends PostgreSQL with distributed capabilities. Most applications never reach the scale where PostgreSQL becomes the bottleneck.

Database

PostgreSQL

The world's most advanced open-source relational database.

Book a Discovery Call PostgreSQL website

PostgreSQL is a mature, ACID-compliant open-source relational database with 35+ years of active development. Extensions like pgvector (vector similarity search), PostGIS (geospatial), pg_cron (scheduled jobs) and TimescaleDB make it a universal data layer — relational, vector, spatial and time-series in one system. Our default database for every project.

Why we chose it

Why Ascii-Core uses PostgreSQL

1
ACID transactions — data integrity guaranteed even under concurrent writes and system failures
2
pgvector extension enables vector similarity search for RAG systems without a separate vector database
3
Mature ecosystem — Prisma, Drizzle, SQLAlchemy and every major ORM support PostgreSQL natively
4
Self-hostable with full control, or managed via Supabase, Neon, RDS — same SQL, any infrastructure

Ideal use cases

When to use PostgreSQL

Relational data with complex joins and aggregations
AI applications requiring vector similarity search via pgvector
Multi-tenant SaaS with Row Level Security for data isolation
Applications requiring full-text search without a separate search engine
Any project where data integrity and ACID compliance matter

Trade-offs

Honest limitations

We believe in accurate trade-offs. These are the cases where a different tool may be the better choice.

Horizontal sharding at extreme scale requires additional tooling (Citus)
Not ideal for extremely high-frequency simple document writes where Firestore or DynamoDB have structural advantages
Requires more configuration than SQLite for development setups

Services