SAP-C02 Question 46
Single answerA large e-commerce company is running its production workloads on Amazon EC2 instances with an Auto Scaling group behind an Application Load Balancer (ALB). During peak traffic, the company observes high latency for requests due to insufficient EC2 capacity being provisioned in time. The company wants to ensure that their Auto Scaling group can better handle sudden traffic spikes while minimizing infrastructure costs. What is the MOST effective solution to address this issue?
- A
Enable predictive scaling for the Auto Scaling group.
- B
Increase the cooldown period for the Auto Scaling group.
- C
Switch the ALB to a Network Load Balancer for faster request handling.
- D
Pre-provision EC2 instances by setting a higher minimum capacity for the Auto Scaling group.
Show answer and explanation
Correct answer: A
Explanation
Predictive scaling is the best solution for this scenario because it proactively launches EC2 instances based on forecasted traffic patterns, ensuring that sufficient capacity is available during sudden spikes without overprovisioning. This feature optimizes both performance and cost, unlike static solutions such as increasing minimum capacity or delaying scaling actions.
- A. Correct.
Predictive scaling uses machine learning to analyze historical data and forecast future traffic patterns, ensuring EC2 instances are provisioned in advance of anticipated spikes. This is the most effective way to handle sudden traffic surges while optimizing costs.
- B. Incorrect.
Increasing the cooldown period delays the launch of new EC2 instances in response to scaling policies, which would exacerbate the problem of insufficient capacity during peak traffic.
- C. Incorrect.
Switching to a Network Load Balancer (NLB) would not address the issue of insufficient EC2 capacity, as NLBs are optimized for low latency and high throughput but do not directly manage scaling behavior.
- D. Incorrect.
While pre-provisioning by increasing the minimum capacity could reduce latency during spikes, it would result in overprovisioning and higher infrastructure costs during normal traffic periods. This is less cost-efficient compared to predictive scaling.