Enabling accelerated recovery for managing public DNS records

Route 53 accelerated recovery for managing public DNS records is designed to achieve a 60-minute Recovery Time Objective (RTO) in the event of service unavailability in the US East (N. Virginia) Region. When enabled on a Route 53 public hosted zone, you will be able to resume making changes to DNS records in the public hosted zone within approximately 60 minutes after AWS detects that operations in the US East (N. Virginia) Region are impaired.

How accelerated recovery for public DNS records works

When accelerated recovery is enabled, Route 53 will maintain a copy of your public hosted zone in the US West (Oregon) Region. If services in the US East (N. Virginia) Region become unavailable for an extended period, Route 53 will execute failover within 60 minutes, automatically redirecting control plane operations for your accelerated recovery enabled public hosted zones to the US West (Oregon) Region. You can then continue to make DNS changes programmatically via the CLI, SDK, and API. Note that a limited set of API methods will be available during failover. See the "Additional considerations" section for more details. When the region recovers, Route 53 will execute the failback procedure, automatically directing control plane operations back to the US East (N. Virginia) Region.

Resubmitting DNS changes after failover

Under normal conditions, changes to public hosted zones with accelerated recovery enabled will be accepted by the US East (N. Virginia) Region and will then be successfully replicated to the US West (Oregon) Region. However, when service disruption occurs in the US East (N. Virginia) Region, some changes may be accepted by the US East (N. Virginia) Region, but may not be replicated to the US West (Oregon) Region. These in-flight changes are referred to as "stranded changes". Once failover completes, Route 53 recommends that you resubmit stranded changes before resuming your DNS workflows. You can achieve this either by using the API, or by using AWS CloudFormation, which are described below.

Using the API to track and submit DNS changes

If you are using the Route 53 API, AWS CLI, or AWS SDKs to manage DNS records, then you will need to use the ChangeResourceRecordSets API and the GetChange API to submit and track DNS changes, respectively.

When you use the ChangeResourceRecordSets API to make DNS changes, Route 53 returns an identifier (ID) for the change you made (see ChangeInfo for change response object details). You can provide this ID in a subsequent request to the GetChange API and observe the status of the change. DNS changes with INSYNC status have been replicated to the US West (Oregon) Region and propagated to all Route 53 DNS servers. There are no further actions you need to take on these changes before or after a failover. However, during impairment to the US East (N. Virginia) Region, GetChange may return PENDING, indicating your change may not have replicated to the US West (Oregon) Region. If that's the case, when failover completes, GetChange will return NoSuchChange, indicating that Route 53 could not replicate this DNS change. Therefore, after failover you can safely disregard these stranded DNS changes and resubmit them as new DNS changes. You will know the failover process has completed when Route 53 posts a message to the AWS Health Dashboard.

Using AWS CloudFormation to track and submit changes

AWS CloudFormation automatically tracks replication status for your DNS changes utilizing the GetChange API, and only completes an update after DNS changes are confirmed as INSYNC. If you are using CloudFormation to manage DNS records and the US East (N. Virginia) Region becomes unavailable, then the actions you take using CloudFormation will not complete successfully during the period of unavailability. However, you can simply retry the same actions to allow CloudFormation to resubmit DNS changes, once the Route 53 failover process completes.

Failback to the US East (N. Virginia) Region

Route 53 will fail back control plane operations for your public hosted zone to the US East (N. Virginia) Region once the region recovers. During the failback, you will not need to resubmit your DNS changes, as no stranded changes will be introduced during this process.

Additional considerations

There are a few additional considerations to be aware of related to the accelerated recovery feature:

How to enable accelerated recovery for managing public DNS records

You can enable accelerated recovery for managing public DNS records using the Route 53 console, API, CLI, or SDK. The time it takes to enable accelerated recovery will depend on the size of your public hosted zone and other factors. You should plan for the process of enabling accelerated recovery to take up to several hours. You can check on the status of the enablement process in the Accelerated recovery tab of your public hosted zone or via the GetHostedZone API. As the process finalizes, there will be a brief period of time lasting up to several minutes where DNS changes are not accepted. Once completed, DNS changes can proceed as normal.

You can disable accelerated recovery using the same steps above, but instead choosing Disable.

CLI example to enable

aws route53 update-hosted-zone-features --enable-accelerated-recovery --hosted-zone-id Z123456789

CLI example to disable

aws route53 update-hosted-zone-features --no-enable-accelerated-recovery --hosted-zone-id Z123456789