How to set up a ZRS & GRS BCP for Azure postgreSQL and how do we recover the database if the region goes down?

We want to set it up using Terraform pipeline. How do we set it up. How do we set below parameters using Terraform.

ZRS, GRS Backup, recover the database on another region using the GRS Backup.

Hi @Gulavani, Amruta

Thank you for the additional details.

Yes, this can be implemented using Terraform. For Azure Database for PostgreSQL Flexible Server, I would recommend separating the Business Continuity/Disaster Recovery design into two levels:

1. Zone-level protection – ZRS / Zone-redundant HA

For protection against an availability-zone failure within the primary region, you can configure zone-redundant high availability (HA). This provisions a standby server in another availability zone in the same region and provides automatic failover.

In regions that support availability zones, PostgreSQL Flexible Server also uses zone-redundant backup storage where applicable. ZRS protects the backup data across availability zones, but it does not by itself provide a database server in another region.

2. Region-level protection – GRS / Geo-redundant backup

For a complete regional disaster recovery scenario, configure geo-redundant backup when creating the PostgreSQL Flexible Server.

The backup data and transaction logs are asynchronously replicated to the Azure paired region. If the primary region becomes unavailable, you can perform a geo-restore, which creates a new PostgreSQL Flexible Server in the paired region using the latest backup data available there.

Please note that geo-redundant backup is not an automatic failover mechanism. The application connection must be redirected to the newly restored server after the recovery operation. There can also be up to approximately one hour of RPO because the geo-redundant backup is replicated asynchronously.

Terraform approach

In Terraform, the PostgreSQL Flexible Server should be created with the appropriate backup retention and geo-redundant backup configuration. The exact Terraform property names depend on the AzureRM provider version being used, so please ensure you are using a current provider version.

Conceptually, the configuration should include:

resource "azurerm_postgresql_flexible_server" "primary" {
  name                = "postgres-primary"
  resource_group_name = azurerm_resource_group.rg.name
  location            = "Switzerland North"

  # compute / storage configuration

  backup_retention_days        = 35
  geo_redundant_backup_enabled = true

  # Zone-redundant HA configuration
  high_availability {
    mode = "ZoneRedundant"
  }
}

The important point is that geo-redundant backup must be enabled when the server is initially created; it cannot be enabled later on an existing server by simply changing the backup redundancy setting.

Regional disaster recovery flow

The recovery pipeline can follow this sequence:

Primary PostgreSQL (Switzerland North)

↓

Geo-redundant backup

↓

Azure paired region

↓

Regional outage detected

↓

Terraform/CLI automation initiates geo-restore

↓

New PostgreSQL Flexible Server created in paired region

↓

Validate database/application connectivity

↓

Update application connection string / DNS / endpoint

For example, Azure CLI provides the az postgres flexible-server geo-restore operation for restoring a Flexible Server from geo-redundant backup.

One important consideration: if the application requires a lower RPO and faster recovery than geo-restore provides, you should also evaluate a cross-region read replica. A cross-region read replica is asynchronously replicated and can be promoted to a standalone read-write server during a regional failure.

Therefore, for a Tier-3 application, a reasonable starting design would be:

Zone failure: Zone-redundant HA → automatic failover Region failure: Geo-redundant backup → geo-restore to paired region Application recovery: Update application endpoint/connection string to the recovered server

If you can share your current Terraform PostgreSQL Flexible Server configuration and the AzureRM provider version, we can review the configuration and provide the corresponding Terraform changes for ZRS/HA, GRS backup, and the regional recovery workflow.

Reference:

• https://learn.microsoft.com/en-us/azure/postgresql/backup-restore/concepts-geo-disaster-recovery
• https://learn.microsoft.com/en-us/azure/postgresql/backup-restore/concepts-backup-restore

I hope this helps. let me know if you have any questions.