Fire and Forget (Async)

Making things feel fast by doing the slow parts in the background.

The idea

When a user signs up for an app, you need to save them to the database and send a welcome email. Saving to the DB takes 10ms. Sending an email via SendGrid takes 800ms. If you do them synchronously, the user waits 810ms staring at a loading spinner. With Fire and Forget, you save to the DB, tell a background system to send the email, and instantly return a success message to the user in 10ms.

How it works (Background Tasks)

In most web frameworks, you cannot just launch a thread and forget it—if the server crashes, the task is lost forever. You use a Message Queue (like Redis + Celery/Sidekiq). You push a tiny JSON job to the queue, and a separate worker process reads it and actually sends the email.

# The synchronous way (Slow UX)
def sign_up():
    db.save_user(email)
    send_email(email) # ⚠ Blocks for 800ms
    return "Welcome!"

# The Fire and Forget way (Fast UX)
def sign_up():
    db.save_user(email)
    
    # ⚡ Pushes {"task":"email", "to": email} to Redis in 1ms
    send_email.delay(email) 
    
    return "Welcome!" # User gets response in 11ms

Cost

This drastically improves perceived latency (UX) and prevents web servers from being tied up waiting on third-party APIs. The trade-off is infrastructural complexity: you now have to deploy, monitor, and scale Redis and a fleet of worker processes.

Watch out for

• Silent Failures: "Forget" doesn't mean you don't care if it fails. If SendGrid is down, the background worker will crash. You MUST configure your worker (e.g. Celery) to automatically retry failed jobs with exponential backoff, and eventually send them to a Dead Letter Queue for manual review.