> ## Documentation Index
> Fetch the complete documentation index at: https://ilanaorm.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Supabase

> Use IlanaORM with a Supabase PostgreSQL database

## Overview

Supabase is a hosted PostgreSQL platform. Because IlanaORM connects directly to PostgreSQL using the `pg` driver, it works with Supabase with no special setup — you just point the connection at your Supabase database URL.

All IlanaORM features work: models, relations, migrations, soft deletes, transactions, and vector search (Supabase has pgvector built in).

***

## Connection setup

Find your connection string in the Supabase dashboard under **Project Settings → Database → Connection string (URI)**. Add it to your `.env` file:

```
DATABASE_URL=postgresql://postgres:[password]@[host]:5432/postgres
```

Then configure IlanaORM:

```javascript theme={null}
// ilana.config.js
const { Database } = require('ilana-orm');

Database.configure({
  default: 'pg',
  connections: {
    pg: {
      client: 'pg',
      connection: {
        connectionString: process.env.DATABASE_URL,
        ssl: { rejectUnauthorized: false },
      },
    },
  },
});
```

<Note>
  The `ssl: { rejectUnauthorized: false }` option is required for Supabase connections. Without it, the connection will be refused.
</Note>

***

## Running migrations

Migrations work exactly the same:

```bash theme={null}
npx ilana migrate
```

IlanaORM will create and manage tables in your Supabase database.

***

## Vector search with Supabase

Supabase includes the `pgvector` extension by default, so IlanaORM's `nearestTo()` and `search()` methods work out of the box. You don't even need to call `enableVectorExtension()` — just configure your model and start querying.

See [Vector Search](/advanced/vector-search) for the full setup.

***

## Serverless deployments (Vercel, Netlify)

In a regular Node.js server, database connections are created once and reused. In serverless environments, each function invocation may spin up a fresh process — creating a new connection every time. If thousands of requests come in, you'd open thousands of connections simultaneously, which overwhelms a regular PostgreSQL database.

Supabase solves this with a **connection pooler** — a proxy that sits in front of the database and manages a shared pool of connections. Instead of connecting directly to your database (port `5432`), you connect to the pooler (port `6543`), and it handles distributing requests across a limited set of real connections.

Get the pooler URL from your Supabase dashboard under **Project Settings → Database → Connection string → Transaction mode**.

```javascript theme={null}
// ilana.config.js — for Vercel / Netlify serverless functions
Database.configure({
  default: 'pg',
  connections: {
    pg: {
      client: 'pg',
      connection: {
        connectionString: process.env.DATABASE_URL, // pooler URL — port 6543
        ssl: { rejectUnauthorized: false },
      },
      pool: { min: 0, max: 1 }, // don't hold open connections between invocations
    },
  },
});
```

<Note>
  `pool: { min: 0, max: 1 }` tells IlanaORM not to keep spare connections alive between requests. In serverless, the process may be frozen or killed after a request anyway, so holding idle connections open just wastes pool slots.
</Note>

***

## Edge runtimes (Cloudflare Workers, Next.js edge routes)

Edge runtimes are more restricted than serverless functions — they have no access to the Node.js file system, so IlanaORM's normal config auto-loading doesn't work. Use the `ilana-orm/edge` entry point, which skips the auto-loader and lets you configure the database manually.

```javascript theme={null}
import { Model, Database } from 'ilana-orm/edge';

Database.configure({
  default: 'pg',
  connections: {
    pg: {
      client: 'pg',
      connection: {
        connectionString: process.env.DATABASE_URL, // pooler URL recommended
        ssl: { rejectUnauthorized: false },
      },
    },
  },
});

// Models and queries work exactly the same
const users = await User.where('is_active', true).get();
```

See [Edge Runtime](/advanced/edge-runtime) for full details and Cloudflare Workers examples.