SAP-C02 Question 279
Select 3Your company has an ecommerce platform running on Amazon EC2 instances behind an Application Load Balancer (ALB). During flash sales, traffic spikes significantly, causing high CPU utilization on the EC2 instances and degraded performance. You have an Auto Scaling group configured, but you notice that scaling out is not happening quickly enough to handle the traffic. Which of the following changes would improve the responsiveness of your Auto Scaling group's scaling policy?
- A
Use a step scaling policy with smaller increments to scale out faster.
- B
Set up a target tracking scaling policy based on the average CPU utilization.
- C
Configure a scheduled scaling policy to scale out in anticipation of the flash sales.
- D
Increase the cooldown period to avoid over-provisioning instances.
- E
Use predictive scaling to forecast demand and scale out ahead of traffic spikes.
Show answer and explanation
Correct answers: B, C, E
Explanation
To address the issue of slow scaling during flash sales, you need to adjust your Auto Scaling group to respond more quickly or preemptively. Target tracking scaling policies automatically adjust scaling based on metrics like CPU utilization, ensuring faster responses. Scheduled scaling is useful when traffic patterns are predictable, such as during flash sales, as it allows you to scale out in advance. Predictive scaling leverages machine learning to forecast demand and scale out ahead of time, making it another effective option for this scenario. Step scaling policies and increasing cooldown periods are less effective solutions in this case.
- A. Incorrect.
Step scaling policies involve scaling in or out based on specific thresholds, but they may not react quickly enough to handle sudden traffic spikes. Smaller increments alone do not guarantee faster scaling.
- B. Correct.
Target tracking scaling policies automatically adjust the capacity of your Auto Scaling group to maintain a specific target metric, such as average CPU utilization. This can result in faster scaling responses.
- C. Correct.
Scheduled scaling allows you to preemptively scale out based on known traffic patterns, such as flash sales, ensuring that instances are ready before the traffic spikes.
- D. Incorrect.
Increasing the cooldown period delays the time between scaling activities, which could make scaling out even slower during traffic spikes. This would worsen the problem rather than fix it.
- E. Correct.
Predictive scaling uses machine learning to forecast traffic and adjust Auto Scaling group capacity proactively. This is ideal for handling predictable traffic spikes like flash sales.