After a lift-and-shift, a web tier runs in one Availability Zone and keeps each user's session on a particular server through load balancer stickiness. The team must improve availability first. Which change should come first?
Show answer and explanation
Answer: C
Spreading instances across several AZs with Auto Scaling behind an Application Load Balancer removes the single-AZ failure point and lets failed instances be replaced automatically. Moving session data into ElastiCache or DynamoDB means any instance can serve any user, so stickiness and dependence on one server are no longer needed. Disabling health checks would keep failed instances in service. Spot capacity in one AZ can be reclaimed at any time, and that AZ remains a single point of failure, so availability would get worse.
Why the other options are wrong
- A. A single AZ is a single point of failure, and stickiness ties users to particular servers. Keeping both leaves the availability risk the team must reduce.
- B. Health checks are what let the load balancer stop routing to failed instances. Disabling them sends users to servers that are not working.
- D. Spot Instances can be interrupted when capacity is reclaimed, and the single AZ stays a single point of failure. Availability would decline rather than improve.