Cloudflare Hyperdrive

Hyperdrive creates a globally-distributed connection pool between Cloudflare Workers and external databases. Workers open a connection to the nearest Hyperdrive node (fast), which maintains a pool of connections to your database (amortised).

Requires: Workers Paid plan. Pricing: $0.15/GB data proxied beyond free tier.

Setup

wrangler hyperdrive create my-hyperdrive \
  --connection-string="postgresql://user:pass@db.example.com:5432/mydb"
# → prints the Hyperdrive ID

wrangler.toml

[[hyperdrive]]
binding = "DB"
id = "YOUR_HYPERDRIVE_ID"

Querying (use standard Postgres driver)

import { Pool } from 'pg';

export default {
  async fetch(request: Request, env: Env): Promise<Response> {
    // Hyperdrive provides a connectionString that routes through the pool
    const pool = new Pool({ connectionString: env.DB.connectionString });

    const { rows } = await pool.query('SELECT * FROM users WHERE active = $1', [true]);
    await pool.end();

    return Response.json(rows);
  },
};

Common Gotchas

• Always close the pool: Call pool.end() at the end of each request — Workers don't persist connection pools between requests.
• Use env.DB.connectionString: This is the Hyperdrive-proxied URL, not your raw database URL.
• SSL is handled by Hyperdrive: No need to configure SSL in the pool — Hyperdrive handles the TLS to your database.
• Hyperdrive is not D1: If you can use D1, use D1. Hyperdrive is for existing external databases you can't migrate.

See Also

• D1 (native edge SQLite — prefer over Hyperdrive when possible)
• Workers runtime