A startup is building a mobile backend API on AWS and wants the system to absorb sudden bursts of millions of requests with no pre-provisioned servers and no dropped requests. Which architecture is most appropriate?
Show answer and explanation
Answer: B
API Gateway and Lambda have no fixed fleet: each request invokes a function, and Lambda raises concurrency automatically to match load within service quotas that can be raised further. API Gateway fronts it with throttling, caching, and request routing, and both services scale to burst traffic in seconds rather than the minutes an instance fleet needs.
With nothing running between requests, there is also no idle cost — a sharp contrast with any pre-provisioned server option.
Why the other options are wrong
- A. EC2 Auto Scaling observes the load metric first and then launches instances, so a sudden burst in the millions can land faster than the fleet can grow, dropping requests during the ramp.
- C. Fargate still runs a scheduled set of containers that must be scaled out task by task, adding scheduling delay for load far beyond what is currently deployed.
- D. Reserved Instances are fixed capacity for steady-state workloads, and the Classic Load Balancer is legacy; neither absorbs unbounded bursty traffic.