You work in an IT unicorn startup as a senior DevOps engineer. Your department is developing an online training platform in AWS. You are responsible for designing a CI/CD pipeline for the said application. One key requirement is that the applications in AWS should be highly available and auto-heal when errors have occurred. Which AWS services can help you achieve this requirement? (Select TWO.)

Answer options:

A.Create an OpsWorks stack with auto-healing feature enabled for its layer.
B.Create an Elastic Beanstalk application with auto-healing feature enabled in the environment configuration.
C.Use a CloudFormation template to launch a stack for both infrastructure and application.
D.Create an Elastic Load Balancer and attach an Auto Scaling Group to it.
E.Create a CloudWatch alarm based on the CPU metrics and notify the on-call team via an SNS subscription.

Correct Answer – A, D
AWS has provided various services to help users deploy applications that are highly available, fault-tolerant, and self-healing. The key for auto-healing is that whenever an error appears in an instance, there is a way to launch a new one while stopping the problematic instance automatically.
Option A is CORRECT: Because OpsWorks has provided an auto-healing feature for its layer. If an agent fails to communicate with the OpsWorks service for some time, AWS OpsWorks Stack automatically replaces the failed instance if the feature is enabled as below:
Option B is incorrect: Because unlike AWS OpsWorks, there is no auto-healing feature for Elastic Beanstalk. Elastic Beanstalk has to rely on ELB and Auto Scaling Group to provide an auto-healing system. Elastic Beanstalk application cannot enable an auto-healing feature.
Option C is incorrect: Although CloudFormation helps create stacks automatically, it does not mention how to achieve auto-healing.
Option D is CORRECT: Because ELB and ASG can help to achieve an elastic and self-healing system very quickly. If any instance goes wrong and fails the health check in ELB, ASG stops the instance and launches a new one to replace it.
Option E is incorrect: Because, although this option can auto alert the on-call team, it still needs manual actions to recover the system.