Improving Auto Scaling Policy

Answer - B and D.

Option A is not necessary as it is handled by CloudWatch by default.

Option B makes sense because maybe the maximum number of servers is low.

Hence the application cannot handle the peak load.

Option C is incorrect because you cannot add servers to an ELB.Option D helps in ensuring Autoscaling can scale the group on the right metrics.

For more information on Autoscaling health checks, please refer to the below document link:

Sure, I'd be happy to explain each answer in detail.

A. Push custom metrics to CloudWatch to monitor your CPU and network bandwidth from your servers, which will allow your Auto Scaling policy to have better fine-grain insight.

This answer is suggesting that you push custom metrics to CloudWatch to monitor the CPU and network bandwidth of your servers. By doing so, you will have more fine-grained insight into how your servers are performing during peak hours. This increased visibility will allow you to make more informed decisions about how to scale your application.

For example, if you see that CPU utilization is high during peak hours, you may need to add more servers to your Auto Scaling group to handle the increased load. Alternatively, if you see that network bandwidth is a bottleneck, you may need to optimize your application or infrastructure to reduce network traffic.

B. Increase your Auto Scaling group's maximum number of servers.

This answer is suggesting that you simply increase the maximum number of servers in your Auto Scaling group. While this may temporarily alleviate the response time issues, it is not necessarily the most efficient or cost-effective solution. Adding more servers may lead to higher costs and may not address the root cause of the response time issues.

C. Create a script that runs and monitors your servers. When it detects an anomaly in load, it posts to an Amazon SNS topic that adds more servers to the load balancer.

This answer is suggesting that you create a script that runs and monitors your servers. When the script detects an anomaly in load, it posts to an Amazon SNS topic that adds more servers to the load balancer. This approach is similar to using custom metrics to monitor your servers but is a more manual approach.

D. Push custom metrics to CloudWatch for your application that includes more detailed information about your web application, such as how many requests it is handling and how many requests are waiting to be processed. Set up scaling policies using the metrics.

This answer is similar to answer A, but it suggests pushing more detailed metrics to CloudWatch. By monitoring metrics such as how many requests are being handled and how many are waiting to be processed, you can gain a better understanding of your application's performance during peak hours. This increased visibility will allow you to set up more effective scaling policies that can automatically adjust the number of servers in your Auto Scaling group based on the application's performance.

In summary, the best answer would be D - pushing custom metrics to CloudWatch that provide more detailed information about your web application and setting up scaling policies based on those metrics. This approach provides the most fine-grained insight into your application's performance during peak hours and allows you to automate scaling decisions based on that performance.