High Availability in Amazon Aurora Relational Database Service

High Availability (HA) for Aurora Relational Database Service (RDS) can be enabled and used by following application modules:

• Keycloak
• Jira
• Confluence
• Mattermost
• SonarQube
• Nexus
• Console

For these modules, you can add more than one RDS instance into a cluster and override variables for it, such as your Availability Zone (AZ). Additional instances will be Reader instances only. If the main Writer instance goes down, Aurora will promote a Reader instance to Writer.

You do not have to specify AZ for each instance; Aurora will automatically place each instance into a different AZ. Side Note: On the back end, Aurora volumes are multi-AZ.

For example, to enable Aurora RDS HA for Keycloak in your env.hcl:

locals {
  keycloak_inputs = {
    rds_instances = {
      primary     = {}
      secondary   = {}
      replica1    = {}
      # ...
    } # Allows a specific number of databases to be defined
  }
}

You can also add auto scaling to the module's HA. Aurora auto scaling will NOT auto scale any instance created by rds_instances in or out; it will only add reader instances based on a count or CPU load of current instances (70%).

Auto scaling is disabled by default. For example, to enable Aurora RDS HA auto scaling for Keycloak in your env.hcl:

locals {
  keycloak_inputs = {
    rds_auto_scale = {
      enabled = true
      min = 0 # default is 0
      max = 5 # default is 5
    }
  }
}