AWS Will Now Let You Reserve Database Capacity Before You Actually Need It

On-Demand Capacity Reservations bring EC2’s “guaranteed slot” model to Aurora and RDS — and you pay On-Demand rates whether you use the reservation or not

Overview

On October 21, 2026, AWS announced On-Demand Capacity Reservations for Amazon Aurora and Amazon RDS — a way to reserve DB instance compute capacity ahead of time so that create, restore, scale-compute, and start operations don’t fail because the Availability Zone you need simply doesn’t have room. It’s the same idea EC2 has offered for years, now applied to the thing that’s historically hardest to improvise around in an emergency: database capacity.

What actually changed

Before this, if you needed to restore a large Aurora cluster during an incident, scale a Multi-AZ RDS deployment up during a traffic spike, or stand up disaster-recovery capacity in a specific zone, you were trusting that AWS had the instance class and capacity you needed available in that zone at that moment. Usually it does. Occasionally — during a regional event, a popular instance-family shortage, or your own badly-timed scaling request — it doesn’t, and the operation that was supposed to save you during an incident fails with an insufficient-capacity error instead.