AWS Auto Scaling

A mechanism that automatically adjusts the number of servers to match load.

A mechanism that increases or decreases the number of resources such as EC2 based on demand. Used for availability and cost optimization.

• Automatic scaling of web server counts
• Replacing failed instances
• Scaling out during access spikes

• Adjusts the number of resources according to load
• Helps with high availability and cost optimization
• Combined with ELB

• Set minimum, desired, and maximum capacity
• Scale based on CloudWatch metrics
• Use Auto Scaling Groups spanning multiple AZs

AWS Auto Scaling is a mechanism that automatically adjusts the number of resources in response to demand. The most widely used form is EC2 Auto Scaling (Auto Scaling Group), which uses CloudWatch metrics (CPU utilization, request count, custom metrics, etc.) as triggers to scale EC2 instances within a minimum, desired, and maximum capacity range. In practice, standard adoption patterns include placing Auto Scaling Groups behind an ALB to provide load distribution and automatic scaling for the web tier, reducing instance counts at night or on weekends to lower costs, and implementing self-healing configurations that automatically replace failed instances. Beyond EC2, the AWS Auto Scaling brand provides cross-resource scaling management (AWS Auto Scaling Plans) for ECS services, DynamoDB read/write capacity, Aurora Read Replicas, and other resource types. Scaling policies include target tracking scaling (deciding scale changes based on metric thresholds), step scaling (incremental adjustments), and scheduled scaling (based on time of day or schedules). Auto Scaling and ELB (ALB / NLB) are designed together as a fundamental pair: ELB handles health checks and traffic distribution, and Auto Scaling handles capacity adjustments. CLF-C02 asks questions such as "Which mechanism automatically scales EC2 instances based on load?" and "Which service is combined with Auto Scaling to achieve high availability? (→ ELB)". Distractor options often plant unrelated services such as "a service that takes storage backups" or "a service that provides DNS" — services from entirely different domains than scaling. Starting at SAA-C03, settings for minimum, desired, and maximum capacity, the design of Auto Scaling Groups spanning multiple AZs, and the distinction between target tracking scaling and step scaling are explored more deeply.

• AWS Auto Scaling (AWS Official)
• What is Amazon EC2 Auto Scaling (AWS Official Documentation)

Last updated: 2026-06-01