Connection Pooling Parameters in Connection Strings

What Is Connection Pooling?

Connection pooling reuses a set of pre-established database connections instead of opening a new connection for every query. This dramatically reduces connection overhead and is essential for production applications that handle concurrent requests.

Why Pooling Matters

Opening a database connection involves TCP handshake, authentication, and TLS negotiation — typically 20-100ms per connection. With pooling:

• A web server handling 100 concurrent requests shares a pool of, say, 20 connections
• Each request borrows a connection, runs its query, and returns it to the pool
• No per-request connection setup overhead

Pool Parameters by Database

PostgreSQL (pgBouncer / driver-level)

postgresql://user:pass@host:5432/db?connection_limit=20&pool_timeout=30

When using an external pooler like PgBouncer:

postgresql://user:pass@pgbouncer-host:6432/db?pgbouncer=true

Prisma-specific pooling with pgbouncer=true disables prepared statements, which are incompatible with PgBouncer's transaction pooling mode.

MySQL

mysql://user:pass@host:3306/db?connectionLimit=10&waitForConnections=true&queueLimit=0

Parameter	Purpose	Default
connectionLimit	Max pool size	10
waitForConnections	Queue when pool full	true
queueLimit	Max queue size (0 = unlimited)	0
idleTimeout	Close idle connections after (ms)	60000

MongoDB

mongodb://user:pass@host:27017/db?maxPoolSize=50&minPoolSize=5&maxIdleTimeMS=30000

Parameter	Purpose	Default
maxPoolSize	Max connections per host	100
minPoolSize	Min idle connections	0
maxIdleTimeMS	Idle connection timeout	0 (no limit)
waitQueueTimeoutMS	Max wait for connection	0 (no limit)

Sizing Guidelines

A common formula for pool size:

pool_size = (core_count * 2) + effective_spindle_count

For a typical 4-core server with SSD storage, this suggests a pool size of about 10. PostgreSQL's documentation recommends keeping the total connection count (across all application instances) below max_connections, which defaults to 100.

Too many connections waste memory (each PostgreSQL connection uses about 10MB of RAM), while too few cause request queuing. Start conservative and scale based on monitoring.