How-to setup a HA/DR database in AWS? [5 - DR database]

In this part of the serie, we will see how to (finally) create a database with a system of disaster recovery.

---

Create a global database

Now that we have a cluster of databases, we can easily create multiple of them in multiple regions and link them to a Global Database.

So at least, we need to create 2 clusters in 2 regions and we will create an aws_rds_global_cluster.

Definition of aws_rds_global_cluster

resource "aws_rds_global_cluster" "example" {
  global_cluster_identifier = "global-test"
  engine                    = "aurora"
  engine_version            = "5.6.mysql_aurora.1.22.2"
  database_name             = "example_db"
}

In this example, we can see that some common parameters are defined here as the engine, its version or the db name to keep an uniformity among all the databases.

---

Updates in the clusters definitions

Due to the fact that you will use a global database, you have to link your clusters to the global one.

global_cluster_identifier = aws_rds_global_cluster.example.id

Then, on all the secondary cluster, you can add depends_on to be sure that they will be created after the global one and the primary one.

depends_on = [
    aws_rds_global_cluster.example,
    aws_rds_cluster.default
  ]

Also, some parameters are not required anymore (because they are now defined in the global database) like :

• master_username
• master_password
• database_name

Then, if you wanted to created all the elements in a single Terraform script, you will have to declare multiple providers for each region where you want to create a cluster. (The complete point will be explain in a following post)

---

Using the Global cluster

Access to the database

The global cluster don't have a specific endpoint to expose the database with Read/Write rights and another for all the read-only databases.

Each cluster will create its own endpoints.

For the primary cluster, both endpoints will work well.

For the other cluster, only the read-only endpoint will work.

It's normal, it's to be sure to have only one main entry point and replicate the update on all the other databases.

But a region is not accessible anymore or if you manually do a failover to change the primary region, the read/write endpoint of the first region will be disabled and the one of the new primary region will be enabled.

How to do manually a failover?

In the AWS web console, select your global database and in the actions you will have Fail over global database. Click on it and the switch will go on!

Links

• Terraform documentation : https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/rds_global_cluster
• AWS failover global database documentation : https://aws.amazon.com/blogs/database/managed-planned-failovers-with-amazon-aurora-global-database/

---

I hope it will help you! 🍺

And see you soon for the next part of this serie. 😀

---

Serie link

• 1 - Start : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-1-1ko7
• 2 - Definitions : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-2-definitions-93p
• 3 - Simple database : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-3-simple-database-a9o
• 4 - HA Database : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-4-ha-database-4kek
• 5 - DR database : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-5-dr-database-278b
• 6 - Create from snapshot : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-6-create-from-snapshot-2mbf
• 7 - Dynamic Terraform backend definition : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-7-dynamic-terraform-backend-definition-3aga
• 8 - Multiple instances in multiple regions : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-8-multiple-instances-in-multiple-regions-210d
• 9 - Generate a random value : https://dev.to/adaendra/how-to-setup-a-hadr-database-in-aws-9-generate-a-random-value-5g8a

---

You want to support me?