3.1. Scalability and Elasticity

💡 First Principle: The cost of over-provisioning for peak load is paid every minute of every day. Elasticity — the ability to scale resources up when demand grows and scale them back down when it falls — is like a rubber band rather than a rigid rod: it stretches under load and returns to rest when the load passes, unlike a fixed allocation that either wastes money at low utilization or fails under peak demand — is how AWS lets you pay only for what you actually use, while still handling any traffic level.

Without elasticity, you have two bad options: provision for peak (wasteful 90% of the time) or provision for average (overloaded during peaks, which is when reliability matters most). Auto Scaling solves this by continuously adjusting capacity to match actual demand.

The exam tests elasticity across three layers of the stack: compute (EC2 Auto Scaling), caching (ElastiCache, CloudFront), and data (RDS, DynamoDB). Getting one layer right but ignoring the others produces a system that scales partially — and fails at its bottleneck.