Why dependency injection matters in FastAPI - Performance Evidence

Performance: Why dependency injection matters

MEDIUM IMPACT

Dependency injection affects how efficiently components are created and reused, impacting server response time and memory usage.

Managing shared resources like database connections in FastAPI endpoints

FastAPI

from fastapi import Depends

def get_db():
    db = DatabaseConnection()
    try:
        yield db
    finally:
        db.close()

def get_user(db: DatabaseConnection = Depends(get_db)):
    return db.query_user()

Reuses the database connection lifecycle efficiently with dependency injection, reducing overhead.

📈 Performance Gainreduces response time by avoiding repeated connection setup, improving throughput

Managing shared resources like database connections in FastAPI endpoints

FastAPI

def get_user():
    db = DatabaseConnection()
    user = db.query_user()
    db.close()
    return user

Creates a new database connection on every request, causing overhead and slower responses.

📉 Performance Costblocks response for extra milliseconds per request due to repeated connection setup

Performance Comparison

Pattern	Object Creation	Resource Usage	Response Time Impact	Verdict
No Dependency Injection	Creates new objects per request	High due to repeated setup	Slower responses	[X] Bad
With Dependency Injection	Reuses objects when possible	Lower resource usage	Faster responses	[OK] Good

Rendering Pipeline

Dependency injection affects the server-side request handling pipeline by controlling object creation and lifecycle, which influences response generation speed.

→Request Handling

→Dependency Resolution

→Response Generation

⚠️ BottleneckRepeated creation of heavy dependencies during request handling

Optimization Tips

1Reuse expensive resources by injecting dependencies instead of creating them per request.

2Avoid repeated setup of connections or services to reduce response time.

3Use FastAPI's Depends system to manage lifecycle and scope of dependencies efficiently.

Performance Quiz - 3 Questions

How does dependency injection improve FastAPI endpoint performance?

A By increasing the number of database connections per request

B By reusing dependencies and reducing repeated object creation

C By delaying response generation until all dependencies are created

D By adding more middleware layers to the request pipeline

DevTools: Performance

How to check: Use a profiling tool or FastAPI middleware to measure request duration and resource usage with and without dependency injection.

What to look for: Look for reduced average request time and lower CPU/memory usage when dependency injection is used.

Practice

1. Why is dependency injection important in FastAPI applications?

easy

A. It forces you to write all code inside one big function.

B. It automatically generates HTML pages for your API.

C. It speeds up the server by caching all responses.

D. It helps keep code clean and makes components easy to share or replace.

Solution

1. Step 1: Understand the purpose of dependency injection

   Dependency injection allows you to provide parts like services or database connections to your functions without hardcoding them.

2. Step 2: Recognize benefits in FastAPI

   This makes your code cleaner and more flexible, as you can easily swap or share components.

3. Final Answer:

   It helps keep code clean and makes components easy to share or replace. -> Option D

4. Quick Check:

   Dependency injection = clean, flexible code [OK]

Hint: Think: clean code and easy swapping [OK]

Common Mistakes:

• Confusing dependency injection with caching
• Thinking it generates HTML automatically
• Believing it forces monolithic code

2. Which of the following is the correct way to declare a dependency in a FastAPI path operation?

easy

A. def read_data(db = Depends(get_db)):

B. def read_data(db: Depends = get_db):

C. def read_data(db: Depends(get_db)):

D. def read_data(db = get_db()):

Solution

1. Step 1: Recall FastAPI dependency syntax

   FastAPI uses Depends() inside the function parameter default value to declare dependencies.

2. Step 2: Check each option

   def read_data(db = Depends(get_db)): correctly uses db = Depends(get_db). Others misuse type hints or call the function directly.

3. Final Answer:

   def read_data(db = Depends(get_db)): -> Option A

4. Quick Check:

   Depends() in default value = correct syntax [OK]

Hint: Use Depends() as default parameter value [OK]

Common Mistakes:

• Calling the dependency function instead of passing it
• Using Depends as a type hint incorrectly
• Assigning dependency without Depends()

3. Given this FastAPI code snippet, what will be printed when accessing the endpoint?

from fastapi import FastAPI, Depends
app = FastAPI()

def get_number():
    return 42

@app.get("/number")
def read_number(num: int = Depends(get_number)):
    print(f"Number is {num}")
    return {"number": num}

medium

A. Number is 42 printed in console, response JSON {"number": 42}

B. Number is 0 printed in console, response JSON {"number": 0}

C. Error because get_number is not awaited

D. Response JSON {"number": null} with no print

Solution

1. Step 1: Understand dependency injection behavior

   The get_number function returns 42 and is injected as the parameter num.

2. Step 2: Trace the endpoint execution

   When the endpoint is called, it prints "Number is 42" and returns JSON with number 42.

3. Final Answer:

   Number is 42 printed in console, response JSON {"number": 42} -> Option A

4. Quick Check:

   Depends injects 42, prints and returns it [OK]

Hint: Dependency returns 42, so print shows 42 [OK]

Common Mistakes:

• Thinking dependency must be async
• Expecting default 0 instead of injected value
• Confusing print output with response content

4. What is wrong with this FastAPI dependency usage?

def get_db():
    return "db_connection"

@app.get("/items")
def read_items(db = get_db()):
    return {"db": db}

medium

A. The function get_db is missing async keyword.

B. The dependency function is called immediately instead of injected.

C. The endpoint is missing a return statement.

D. Depends() is used incorrectly inside the function body.

Solution

1. Step 1: Identify how dependency should be declared

   Dependencies must be passed as Depends(get_db), not by calling get_db() directly.

2. Step 2: Explain the problem in the code

   Calling get_db() runs it once at startup, not per request, losing flexibility and benefits of injection.

3. Final Answer:

   The dependency function is called immediately instead of injected. -> Option B

4. Quick Check:

   Call vs Depends() matters for injection [OK]

Hint: Use Depends(), don't call dependency function [OK]

Common Mistakes:

• Calling dependency function instead of passing Depends()
• Confusing async requirement with dependency injection
• Ignoring missing return statement (it exists here)

5. You want to share a database connection across multiple endpoints in FastAPI using dependency injection. Which approach best ensures the connection is created once per request and properly closed after?

hard

A. Pass the connection as a query parameter to each endpoint.

B. Create the connection globally once outside any function and reuse it everywhere.

C. Use a dependency function with yield that opens the connection, yields it, then closes it after.

D. Call the connection function directly inside each endpoint without Depends.

Solution

1. Step 1: Understand lifecycle management with dependencies

   Using yield in a dependency function allows setup before yield and cleanup after the request finishes.

2. Step 2: Compare options for connection management

   Use a dependency function with yield that opens the connection, yields it, then closes it after. correctly manages connection per request lifecycle. Others either create global state or misuse parameters.

3. Final Answer:

   Use a dependency function with yield that opens the connection, yields it, then closes it after. -> Option C

4. Quick Check:

   Yield in dependency = setup and cleanup per request [OK]

Hint: Use yield in dependency for setup and cleanup [OK]

Common Mistakes:

• Using global connection risking concurrency issues
• Calling connection directly losing cleanup control
• Passing connection via query parameters insecurely