Skip to content

python.io_in_hot_path

Performance Medium

Detects I/O operations (file access, network calls, database queries) in hot code paths like loops or frequently-called functions.

Why It Matters

I/O operations in hot paths cause:

Latency spikes — Each I/O call adds milliseconds of delay
Throughput reduction — Blocking I/O prevents concurrent processing
Resource exhaustion — Too many concurrent I/O operations overwhelm systems
Unpredictable performance — I/O latency varies significantly

Example

# ❌ Before (I/O in loop)
def process_users(user_ids):
    results = []
    for user_id in user_ids:
        # I/O call in every iteration
        user = fetch_user_from_db(user_id)
        results.append(transform(user))
    return results

# ✅ After (batch I/O)
def process_users(user_ids):
    # Single I/O call for all users
    users = fetch_users_batch(user_ids)
    return [transform(user) for user in users]

What Unfault Detects

Database queries inside loops
HTTP requests inside loops
File reads/writes inside loops
Network operations in frequently-called functions
I/O in list comprehensions or generator expressions

Auto-Fix

Unfault generates patches that suggest batching strategies:

# Batch database queries
users = User.query.filter(User.id.in_(user_ids)).all()
user_map = {u.id: u for u in users}

# Batch HTTP requests
import asyncio
results = await asyncio.gather(*[fetch(url) for url in urls])

Performance Comparison

Pattern	100 items	1000 items
I/O in loop	~500ms	~5000ms
Batched I/O	~10ms	~50ms