Building a high available Anycast service using AWS global accelerator

This is a terraform recipe for spinning up an AWS Global Accelerator instance.

Also make sure to check the accompanying blog post here: Building a high available Anycast service using AWS global accelerator

This accelerator will have two listeners, one for UDP53 and one for TCP80. As part of the Terraform recipe, we also spin up vpc's in 4 regions, each with two EC2 instances. This repo comes with a DNS and Webserver (GoLang) that will be installed on the EC2 instances (see /scripts/ )

To get started: Make sure have your AWS credentials ready.

Also make sure the ssh key path is set correctly, currently hardcoded as ~/.ssh/id_rsa.pub in vpc/010-ssh-key.tf

To start the terraform work, simply type:

terraform init 
terraform plan 
terraform apply 

The terraform output will show the Global Accelerator IPs. Example output:

Output:
<SNIP>
GlobalAccelerator = [
  [
    [
      {
        "ip_addresses" = [
          "13.248.138.197",
          "76.223.8.146",
        ]
        "ip_family" = "IPv4"
      },
    ],
  ],
]
<SNIP>

To test the web server (make sure the replace the IP with the IP of your GlobalAccelerator):

for i in {1..30}; do curl -s -q  13.248.138.197 ;done |sort | uniq -c | sort -nr
  18 hostname: i-0e32d5b1a3f76cc05.us-west-2c
  12 hostname: i-006b93ffb1844c92f.us-west-2c

To test the DNS server (make sure the replace the IP with the IP of your GlobalAccelerator):

for i in {1..20}; do dig @13.248.138.197 ID.SERVER txt ch +short ; done | sort | uniq -c | sort -nr
  10 "i-0e32d5b1a3f76cc05.us-west-2c"
  10 "i-006b93ffb1844c92f.us-west-2c"