Background

When migrating applications to AWS public cloud, we typically use ELB to achieve application high availability and load balancing. In reality, there are some applications with low load that are published through EIP. If the EC2 where the application is located fails or the availability zone fails, the application becomes unavailable. This article introduces a method to achieve application high availability in the above scenario using AWS services.

Principle

Use EC2's own status check function to implement EC2 termination operation.

Use AutoScaling function to monitor EC2 node count and start new EC2 instances.

Steps

1. Create an IAM role that can execute EC2 CLI commands;
2. Create an EC2 with the IAM role from step 1, deploy the application, and bind an EIP;
3. Create a script on the EC2, add the script execution to the rc.local file to automatically bind EIP on boot, with the following content:

1. After testing without issues, create an AMI image file from the above EC2 (create a new AMI image for each application upgrade);
2. Configure EC2 system status check vs instance status check, when status is abnormal, execute instance termination operation;
3. Create a launch configuration based on the AMI from step 4, with the role from step 1;
4. Use the launch configuration from step 6 to add an Auto Scaling group, specify multiple availability zones and subnets for the subnet, group size is 1;
5. Add the current EC2 to the Auto Scaling group from step 7, configuration complete.

After the above configuration is complete, when EC2 status is abnormal or availability zone status is abnormal, automatic failover at the minute level is achieved for the application, increasing application high availability.