Course contentsShow
← Cloud and Infrastructure
Lesson 293 of 2,9896. Auto ScalingFree lesson

Cooldown Periods

Learning how cooldown periods prevent rapid successive scaling actions and ensure stability.

What You'll Learn

You'll learn how cooldown periods act as a stability buffer, preventing your Auto Scaling system from making too many changes too quickly.

The Problem Without Cooldown

Imagine turning your home thermostat up and down every few seconds because the room didn't instantly reach your target temperature. You'd create chaos and waste energy. Auto Scaling faces the same challenge: when a scaling action triggers (adding or removing instances), it takes time for those changes to fully take effect and for metrics to stabilize.

Without cooldown periods, Auto Scaling might see high CPU, launch new instances, but then immediately see metrics that still look high (because the new instances haven't started handling traffic yet), and launch even more instances. This creates a cascading overreaction.

What is a Cooldown Period?

A cooldown period is a configurable waiting time (typically 300 seconds/5 minutes by default) that prevents Auto Scaling from initiating another scaling action immediately after a previous one completes.

During cooldown:

  • Auto Scaling will not launch or terminate additional instances based on the same alarm
  • Newly launched instances have time to start up and begin handling requests
  • Metrics can stabilize to reflect the true system state
  • Your infrastructure avoids thrashing (rapid, repeated scaling up and down)

How It Works

  1. A scaling policy triggers (e.g., CPU > 70%)
  2. Auto Scaling launches 2 new instances
  3. Cooldown timer starts (e.g., 300 seconds)
  4. During these 5 minutes, even if CPU is still high, no additional scaling occurs
  5. After cooldown expires, Auto Scaling can respond to current conditions again

Key Takeaway: Cooldown periods are waiting windows that give your infrastructure time to stabilize after scaling actions, preventing rapid, wasteful, and potentially destabilizing successive changes.