Capacity Limits and Constraints
What you'll learn: How to set minimum, maximum, and desired capacity values to keep your auto scaling group operating safely within defined boundaries.
The Control Boundaries
Think of auto scaling capacity limits like the thermostat range in a smart home. You don't want the temperature dropping below 65°F (minimum) or soaring above 78°F (maximum), but you prefer it around 72°F (desired). Auto Scaling Groups work the same way—you define hard boundaries to prevent under-provisioning or runaway costs.
The Three Capacity Values
Minimum Capacity sets the floor—the smallest number of instances that must always run, even during off-peak times. This ensures your application never goes completely offline.
Maximum Capacity sets the ceiling—the upper limit on instances, protecting you from unexpected cost spikes if a scaling policy overreacts to traffic surges.
Desired Capacity is your target under normal conditions. Auto Scaling tries to maintain this number, but it can flex between your minimum and maximum based on scaling policies and current demand.
Why Constraints Matter
Without proper constraints, you risk two extremes: scaling down to zero instances (application outage) or scaling up indefinitely (budget disaster). These boundaries give you:
- Cost control: Maximum capacity prevents surprise bills
- Availability guarantee: Minimum capacity ensures service continuity
- Operational predictability: Desired capacity sets your baseline
When you configure a scaling policy (like target tracking or step scaling), it still respects these limits. If a policy wants to add more instances but you've hit maximum capacity, scaling stops there.
Key Takeaway: Minimum, maximum, and desired capacity values act as safety guardrails for your auto scaling group, ensuring it operates within acceptable cost and availability boundaries regardless of what your scaling policies request.