Achieving High Availability with Aurora Global Databases

For mission-critical applications like FinTech ledgers or global e-commerce sites, a localized region outage cannot be tolerated. Standard RDS cross-region read replicas use asynchronous replication, which can introduce replication lag and slower failover times during a disaster recovery scenario.

Amazon Aurora Global Databases are designed for globally distributed applications, allowing a single Aurora database to span multiple AWS regions. It utilizes dedicated infrastructure to replicate data with typical latency of under a second, completely offloading the replication process from the database engine to the storage layer.

In the event of a primary region degradation, a secondary region can be promoted to full read/write capabilities in under a minute. Setting this up via Terraform involves defining an `aws_rds_global_cluster` resource, and then attaching regional cluster instances to that global identifier.

resource "aws_rds_global_cluster" "global_db" {
  global_cluster_identifier = "fintech-global-ledger"
  engine                    = "aurora-postgresql"
  engine_version            = "13.6"
  database_name             = "ledgerdb"
}

resource "aws_rds_cluster" "primary" {
  provider                  = aws.primary_region
  engine                    = aws_rds_global_cluster.global_db.engine
  engine_version            = aws_rds_global_cluster.global_db.engine_version
  cluster_identifier        = "fintech-primary-cluster"
  global_cluster_identifier = aws_rds_global_cluster.global_db.id
  master_username           = "admin"
  master_password           = random_password.master.result
}