Designing High-Performing Compute Architecture

1. What is the best reason for creating golden AMIs for your web servers and application servers?    
A golden AMI is an image that is as perfect as possible, and no customizations or tweaks need to be made. You need to have the right processes in place to properly test your images before they go into production. Once your application servers are in production, you want to be able to have a “hands off” policy. You do not need to troubleshoot your production servers if your AMIs have been fully tested in preproduction.

2. Can you replace existing third-party load balancers with ELB services?
AWS supports third-party load balancer appliances through the deployment of a gateway load balancer, which enables you to automate the scale of your third-party load balancer in operation.

3. Should your application servers be placed in public or private subnets?
Application service should always be hosted in private subnets, away from direct Internet access. Public-facing load balancers should be hosted in a public subnet to protect the application stack from direct access. All requests for your application have to communicate first with the load balancer, which in turn directs your request to the appropriate target group.

4. What would be a good reason to consider installing the CloudWatch agent?
The CloudWatch agent has probably already been deployed. You might want to consider customizing the operation of the CloudWatch agent so that your application logs and system logs are sent to CloudWatch for analysis. CloudWatch can then send notifications when problems arise.

5. What is the recommended process for planning an Auto Scaling deployment? In this process, should you use a launch configuration or a launch template?
Although launch configurations can be used to create EC2 instances that are performed by the Auto Scaling service as required, launch configurations are being deprecated and replaced with launch template. Launch templates support all features of EC2 instances, and launch configurations do not support all current and new EC2 features.

6. What is the recommended Auto Scaling policy setting to start with?    
The recommended starting point is to deploy a target tracking policy, which allows you to define the level of application performance to be maintained. Auto Scaling adds and removes compute instances to meet this target level.

7. Do simple scaling policies help improve application performance?
Simple scaling policies can help improve application performance. However, only a single metric can be used to determine the amount of scaling up and down. In addition, after a change has been made, a cooldown period must be observed before any additional changes can be performed.

8. Do step scaling policies help you save money?
Step scaling enables you to define the scaling of an application’s compute levels based on multiple percentages when scaling both up and down. Therefore, step scaling allows you to carefully tune your scaling requirements.