Connection pooling in FastAPI

Introduction

Connection pooling helps your FastAPI app reuse database connections. This makes your app faster and uses fewer resources.

When your app talks to a database many times per second.

When you want to avoid the delay of opening a new database connection each time.

When you want to limit how many database connections your app uses at once.

When you want your app to handle many users smoothly without slowing down.

Syntax

FastAPI

from sqlalchemy.ext.asyncio import create_async_engine, AsyncSession
from sqlalchemy.orm import sessionmaker

engine = create_async_engine(
    "postgresql+asyncpg://user:password@localhost/dbname",
    pool_size=5,  # Number of connections to keep in the pool
    max_overflow=10,  # Extra connections allowed beyond pool_size
)

async_session = sessionmaker(
    engine, expire_on_commit=False, class_=AsyncSession
)

Use create_async_engine to create a database engine with connection pooling.

pool_size controls how many connections stay open ready to use.

Examples

This sets a pool of 10 connections and allows 20 extra temporary connections if needed.

FastAPI

engine = create_async_engine(
    "postgresql+asyncpg://user:password@localhost/dbname",
    pool_size=10,
    max_overflow=20
)

This creates a session factory that uses the connection pool to get connections.

FastAPI

async_session = sessionmaker(
    engine, expire_on_commit=False, class_=AsyncSession
)

Sample Program

This FastAPI app uses connection pooling to talk to a SQLite database asynchronously. The get_session function provides a session from the pool. The endpoint runs a simple query and returns the result.

FastAPI

from fastapi import FastAPI, Depends
from sqlalchemy.ext.asyncio import create_async_engine, AsyncSession
from sqlalchemy.orm import sessionmaker
from sqlalchemy import text

app = FastAPI()

DATABASE_URL = "sqlite+aiosqlite:///./test.db"

engine = create_async_engine(
    DATABASE_URL
)

async_session = sessionmaker(
    engine, expire_on_commit=False, class_=AsyncSession
)

async def get_session() -> AsyncSession:
    async with async_session() as session:
        yield session

@app.get("/items")
async def read_items(session: AsyncSession = Depends(get_session)):
    result = await session.execute(text("SELECT 1 as number"))
    number = result.scalar_one()
    return {"number": number}

Important Notes

Connection pools keep connections open so your app can reuse them quickly.

Adjust pool_size and max_overflow based on your app's load and database limits.

Always close sessions properly to return connections to the pool.

Summary

Connection pooling speeds up database access by reusing connections.

FastAPI with SQLAlchemy supports connection pooling easily.

Configure pool size to balance performance and resource use.

Practice

1. What is the main benefit of using connection pooling in FastAPI applications?

easy

A. It caches API responses for faster delivery

B. It reuses database connections to improve performance

C. It automatically creates database tables

D. It encrypts data sent to the database

Solution

1. Step 1: Understand connection pooling purpose

   Connection pooling keeps database connections open and reuses them instead of opening new ones each time.

2. Step 2: Identify the benefit in FastAPI context

   This reuse reduces the time spent opening connections, speeding up database access in FastAPI apps.

3. Final Answer:

   It reuses database connections to improve performance -> Option B

4. Quick Check:

   Connection pooling = reuse connections = better speed [OK]

Hint: Pooling means reusing connections, not creating or encrypting [OK]

Common Mistakes:

• Confusing pooling with encryption
• Thinking pooling creates tables automatically
• Mixing pooling with API response caching

2. Which of the following is the correct way to set the maximum pool size using SQLAlchemy in FastAPI?

easy

A. engine = create_engine(DB_URL, pool_size=10)

B. engine = create_engine(DB_URL, max_connections=10)

C. engine = create_engine(DB_URL, max_pool=10)

D. engine = create_engine(DB_URL, connection_limit=10)

Solution

1. Step 1: Recall SQLAlchemy pool size parameter

   The correct parameter to set max pool size is pool_size.

2. Step 2: Match parameter with options

   Only engine = create_engine(DB_URL, pool_size=10) uses pool_size=10, which is valid syntax.

3. Final Answer:

   engine = create_engine(DB_URL, pool_size=10) -> Option A

4. Quick Check:

   pool_size sets max connections in SQLAlchemy [OK]

Hint: Use pool_size to set max connections in create_engine [OK]

Common Mistakes:

• Using incorrect parameter names like max_connections
• Confusing pool_size with connection_limit
• Trying to set max_pool which is invalid

3. Given this FastAPI code snippet using SQLAlchemy, what will be the output when the app handles multiple requests?

from sqlalchemy import create_engine
engine = create_engine('sqlite:///test.db', pool_size=5, max_overflow=0)

# Each request uses engine.connect()

medium

A. Only one connection is used for all requests

B. Each request creates a new connection ignoring pool size

C. Up to 5 connections are reused; new requests wait if all are busy

D. The app crashes due to too many connections

Solution

1. Step 1: Understand pool_size effect

   Setting pool_size=5 means the engine keeps up to 5 connections open for reuse.

2. Step 2: Behavior on multiple requests

   When more than 5 requests come, new ones wait until a connection is free; no new connections beyond 5 are created.

3. Final Answer:

   Up to 5 connections are reused; new requests wait if all are busy -> Option C

4. Quick Check:

   pool_size=5 means max 5 reusable connections [OK]

Hint: pool_size limits max open connections reused [OK]

Common Mistakes:

• Assuming each request creates a new connection
• Thinking only one connection is shared
• Believing the app crashes when pool is full

4. You wrote this FastAPI code with SQLAlchemy connection pooling:

engine = create_engine('postgresql://user:pass@localhost/db', pool_size=5, max_overflow=0)
connection = engine.connect()
# forgot to close connection after use

What problem will this cause?

medium

A. No problem; connections close automatically

B. The database will reject all connections immediately

C. The app will automatically close connections after timeout

D. Connections will be exhausted causing new requests to hang

Solution

1. Step 1: Identify missing connection close

   Not closing connections means they stay checked out and unavailable for reuse.

2. Step 2: Effect on connection pool

   Since pool_size=5, after 5 connections are checked out and not closed, no free connections remain, causing new requests to wait indefinitely.

3. Final Answer:

   Connections will be exhausted causing new requests to hang -> Option D

4. Quick Check:

   Not closing connections exhausts pool causing hangs [OK]

Hint: Always close connections to free pool slots [OK]

Common Mistakes:

• Assuming connections close automatically without code
• Thinking database rejects connections immediately
• Believing app auto-closes connections after timeout

5. You want to optimize a FastAPI app using SQLAlchemy with a PostgreSQL database. The app has many short requests. Which pooling configuration best balances performance and resource use?

Options:
A) pool_size=20, max_overflow=10
B) pool_size=1, max_overflow=50
C) pool_size=5, max_overflow=0
D) pool_size=0, max_overflow=0

hard

A. pool_size=5, max_overflow=0

B. pool_size=1, max_overflow=50

C. pool_size=20, max_overflow=10

D. pool_size=0, max_overflow=0

Solution

1. Step 1: Understand pool_size and max_overflow roles

   pool_size sets fixed connections; max_overflow allows extra temporary connections beyond pool_size.

2. Step 2: Analyze options for many short requests

   Too large pool_size wastes resources; too small with high overflow risks overhead. pool_size=5, max_overflow=0 with moderate pool_size=5 and no overflow balances reuse and resource use well.

3. Final Answer:

   pool_size=5, max_overflow=0 -> Option A

4. Quick Check:

   Moderate pool_size with no overflow balances performance and resources [OK]

Hint: Moderate pool_size with zero overflow balances well [OK]

Common Mistakes:

• Choosing very high pool_size wasting resources
• Using zero pool_size disables pooling
• Setting high max_overflow causes many temporary connections