Question

A CloudOps engineer manages a stateless web tier running on an EC2 Auto Scaling group backed by a single launch template that specifies one instance type and 100% On-Demand capacity. During a regional capacity event, the group failed to launch replacement instances because the specified instance type was temporarily unavailable in two of the three subnets, causing reduced availability. The team wants to improve the group's resilience to capacity shortages while also lowering cost, without managing multiple Auto Scaling groups. What should the engineer do?

Accepted Answer

Configure a mixed instances policy on the Auto Scaling group with multiple instance types and a combination of On-Demand and Spot capacity, using the capacity-optimized allocation strategy.

Answer

Create three separate Auto Scaling groups, one per subnet, each with a different instance type, and place them behind the same load balancer.

Answer

Increase the desired capacity of the existing group and enable predictive scaling so extra instances are pre-provisioned before capacity shortages occur.

Answer

Switch the launch template to use a larger instance type from the same family and enable detailed monitoring to detect launch failures faster.