Using NoSQL (MongoDB, Redis)

MongoDB with Motor

Motor is the official async MongoDB driver for Python. It uses the same API as pymongo but with async/await.

Setup

pip install motor

# app/database.py
from motor.motor_asyncio import AsyncIOMotorClient

MONGO_URL = "mongodb://localhost:27017"
client: AsyncIOMotorClient | None = None

async def connect_db():
    global client
    client = AsyncIOMotorClient(MONGO_URL)

async def close_db():
    if client:
        client.close()

def get_db():
    return client["techblog"]

Lifespan Integration

# main.py
from contextlib import asynccontextmanager
from fastapi import FastAPI
from app.database import connect_db, close_db

@asynccontextmanager
async def lifespan(app: FastAPI):
    await connect_db()
    yield
    await close_db()

app = FastAPI(lifespan=lifespan)

CRUD Operations

# app/routes.py
from fastapi import APIRouter, HTTPException, status
from app.database import get_db
from bson import ObjectId
from pydantic import BaseModel, Field
from typing import Optional

router = APIRouter(prefix="/items", tags=["items"])


# ── Schemas ──

class ItemCreate(BaseModel):
    name: str
    price: float
    description: Optional[str] = None


class ItemResponse(BaseModel):
    id: str = Field(alias="_id")
    name: str
    price: float
    description: Optional[str] = None


# ── Routes ──

@router.post("/", status_code=status.HTTP_201_CREATED)
async def create_item(payload: ItemCreate):
    db = get_db()
    result = await db.items.insert_one(payload.model_dump())
    created = await db.items.find_one({"_id": result.inserted_id})
    created["_id"] = str(created["_id"])
    return created


@router.get("/{item_id}")
async def get_item(item_id: str):
    db = get_db()
    try:
        obj_id = ObjectId(item_id)
    except Exception:
        raise HTTPException(status_code=400, detail="Invalid ID format")
    item = await db.items.find_one({"_id": obj_id})
    if not item:
        raise HTTPException(status_code=404, detail="Item not found")
    item["_id"] = str(item["_id"])
    return item


@router.get("/")
async def list_items(skip: int = 0, limit: int = 20):
    db = get_db()
    items = []
    cursor = db.items.find().skip(skip).limit(limit)
    async for doc in cursor:
        doc["_id"] = str(doc["_id"])
        items.append(doc)
    return items


@router.put("/{item_id}")
async def update_item(item_id: str, payload: ItemCreate):
    db = get_db()
    try:
        obj_id = ObjectId(item_id)
    except Exception:
        raise HTTPException(status_code=400, detail="Invalid ID format")
    result = await db.items.replace_one({"_id": obj_id}, payload.model_dump())
    if result.matched_count == 0:
        raise HTTPException(status_code=404, detail="Item not found")
    updated = await db.items.find_one({"_id": obj_id})
    updated["_id"] = str(updated["_id"])
    return updated


@router.delete("/{item_id}", status_code=status.HTTP_204_NO_CONTENT)
async def delete_item(item_id: str):
    db = get_db()
    try:
        obj_id = ObjectId(item_id)
    except Exception:
        raise HTTPException(status_code=400, detail="Invalid ID format")
    result = await db.items.delete_one({"_id": obj_id})
    if result.deleted_count == 0:
        raise HTTPException(status_code=404, detail="Item not found")

MongoDB Schema Notes

Aspect	MongoDB Behavior
ID field	_id is an ObjectId - convert to string for JSON
Schema	Schema-less - documents in the same collection can have different fields
Indexes	Create with db.collection.create_index("field")
Relations	Use $lookup aggregation or manual references (no JOINs)
Transactions	Supported since MongoDB 4.0 (requires replica set)

Redis with redis-py

redis-py is the official Python client. Use aioredis (merged into redis-py 4.x+) for async support.

Setup

pip install redis

# app/redis_client.py
from redis.asyncio import Redis

redis_client: Redis | None = None

async def init_redis():
    global redis_client
    redis_client = Redis(host="localhost", port=6379, decode_responses=True)

async def close_redis():
    if redis_client:
        await redis_client.aclose()

def get_redis() -> Redis:
    return redis_client

Lifespan Integration

@asynccontextmanager
async def lifespan(app: FastAPI):
    await connect_db()
    await init_redis()
    yield
    await close_redis()
    await close_db()

Redis CRUD

router = APIRouter(prefix="/sessions", tags=["sessions"])


@router.post("/{session_id}")
async def set_session(session_id: str, data: dict):
    r = get_redis()
    await r.hset(f"session:{session_id}", mapping=data)
    await r.expire(f"session:{session_id}", 3600)  # 1 hour TTL
    return {"status": "ok"}


@router.get("/{session_id}")
async def get_session(session_id: str):
    r = get_redis()
    data = await r.hgetall(f"session:{session_id}")
    if not data:
        raise HTTPException(status_code=404, detail="Session not found")
    return data


@router.delete("/{session_id}", status_code=204)
async def delete_session(session_id: str):
    r = get_redis()
    await r.delete(f"session:{session_id}")

Caching Patterns

Pattern 1: Cache-Aside

@router.get("/items/{item_id}")
async def get_item_with_cache(item_id: str):
    r = get_redis()
    cache_key = f"item:{item_id}"

    # Try cache first
    cached = await r.get(cache_key)
    if cached:
        return {"source": "cache", "data": json.loads(cached)}

    # Fall through to DB
    db = get_db()
    try:
        obj_id = ObjectId(item_id)
    except Exception:
        raise HTTPException(status_code=400, detail="Invalid ID")
    item = await db.items.find_one({"_id": obj_id})
    if not item:
        raise HTTPException(status_code=404, detail="Item not found")
    item["_id"] = str(item["_id"])

    # Write to cache
    await r.setex(cache_key, 300, json.dumps(item, default=str))
    return {"source": "database", "data": item}

Pattern 2: Invalidation on Write

@router.put("/items/{item_id}")
async def update_item_clear_cache(item_id: str, payload: ItemCreate):
    db = get_db()
    r = get_redis()
    try:
        obj_id = ObjectId(item_id)
    except Exception:
        raise HTTPException(status_code=400, detail="Invalid ID")
    result = await db.items.replace_one({"_id": obj_id}, payload.model_dump())
    if result.matched_count == 0:
        raise HTTPException(status_code=404, detail="Item not found")
    # Invalidate cache
    await r.delete(f"item:{item_id}")
    updated = await db.items.find_one({"_id": obj_id})
    updated["_id"] = str(updated["_id"])
    return updated

Pattern 3: Rate Limiting

from fastapi import Request

@app.middleware("http")
async def rate_limit_middleware(request: Request, call_next):
    r = get_redis()
    client_ip = request.client.host
    key = f"ratelimit:{client_ip}"

    count = await r.incr(key)
    if count == 1:
        await r.expire(key, 60)  # 1 minute window

    if count > 100:
        return JSONResponse(status_code=429, content={"detail": "Too many requests"})

    return await call_next(request)

Redis Command Reference

Command	Python	Use Case
SET key value [EX seconds]	await r.set(key, val, ex=ttl)	String cache
GET key	await r.get(key)	Read cache
HSET key field value	await r.hset(key, mapping={})	Object store
HGETALL key	await r.hgetall(key)	Read object
EXPIRE key seconds	await r.expire(key, ttl)	Set TTL
INCR key	await r.incr(key)	Counters
DEL key	await r.delete(key)	Remove key
PUBLISH channel msg	await r.publish(ch, msg)	Pub/Sub

SQL vs NoSQL - How to Choose

Factor	SQL (PostgreSQL)	NoSQL (MongoDB)	Redis
Schema	Rigid, enforced	Flexible, no enforcement	Key-value / struct
Relations	JOINs, foreign keys	Embedded docs, $lookup	Manual
ACID	Full ACID	Multi-doc ACID (4.0+)	No (single key atomic)
Query	Complex joins, aggregations	Rich query language	By key only
Scaling	Vertical (read replicas)	Horizontal (sharding)	Horizontal (cluster)
Best for	Relational data, reporting	Rapid prototyping, varied docs	Caching, sessions, queues
FastAPI fit	SQLAlchemy + Alembic	Motor + FastAPI	redis-py + Depends

Key Takeaways

• Use Motor (motor.motor_asyncio) for async MongoDB - it mirrors pymongo with await on every call
• Convert ObjectId to str before returning JSON responses; use bson.objectid.ObjectId for queries
• Use redis-py (Redis.asyncio) for Redis - hset/hgetall for objects, setex for TTL caches
• Implement cache-aside pattern: check Redis first, fall back to DB, write to Redis on miss
• Invalidate cache on writes to prevent stale data; use EXPIRE/setex for automatic eviction
• Choose SQL for complex queries and relationships; MongoDB for flexible schemas; Redis for caching and ephemeral data