Best practices for routing control in Route 53 ARC

In a disaster recovery (DR) scenario, keep system dependencies to a minimum by using a simple approach to accessing Amazon and performing recovery tasks. Create IAM long-lived credentials specifically for DR tasks, and keep the credentials securely in an on-premises physical safe or a virtual vault, to access when needed. With IAM, you can centrally manage security credentials, such as access keys, and permissions for access to Amazon resources. For non-DR tasks, we recommend that you continue to use federated access, using Amazon services such as Amazon Single Sign-On.

To perform failover tasks in Route 53 ARC with the recovery cluster data plane API, you can attach a Route 53 ARC IAM policy to your user. To learn more, see Identity-based policy examples in Amazon Route 53 Application Recovery Controller.

For DNS records that you might need to change as part of your failover mechanism, especially records that are health checked, using lower TTL values is appropriate. Setting a TTL of 60 or 120 seconds is a common choice for this scenario.

The DNS TTL (time to live) setting tells DNS resolvers how long to cache a record before requesting a new one. When you choose a TTL, you make a trade-off between latency and reliability, and responsiveness to change. With a shorter TTL on a record, DNS resolvers notice updates to the record more quickly because the TTL specifies that they must query more frequently.

For more information, see Choosing TTL values for DNS records in Best practices for Amazon Route 53 DNS.

We recommend that you keep a local copy of your Route 53 ARC Regional cluster endpoints, in bookmarks or saved in automation code that you use to retry your endpoints. During a failure event, you might not be able to access some API operations, including Route 53 ARC API operations that are not hosted on the extremely reliable data plane cluster. You can list the endpoints for your Route 53 ARC clusters by using the DescribeCluster API operation.

We recommend that when you need to fail over, you update (and retrieve) routing control states using a random endpoint from your five Regional cluster endpoints. If that endpoint fails, then retry each of your other Regional endpoints. For information about using code examples with the Amazon SDK, including examples for trying cluster endpoints, see Code examples for Application Recovery Controller using Amazon SDKs.

Using the Route 53 ARC data plane API, view your routing controls and states with the ListRoutingControls operation and update routing control states to redirect traffic for failover with the UpdateRoutingControlState operation. You can use the Amazon CLI (as in these examples) or code that you write using one of the Amazon SDKs. Route 53 ARC offers extreme reliability with the API in the data plane to fail over traffic. We recommend using the API instead of changing routing control states in the Amazon Web Services Management Console.

Connect to one of your Regional cluster endpoints for Route 53 ARC to use the data plane API. If the endpoint is unavailable, try connecting to another cluster endpoint.

If a safety rule blocks a routing control state update, you can bypass it to make the update and fail over traffic. For more information, see Overriding safety rules to reroute traffic.

Test failover regularly with Route 53 ARC routing control, to fail over from your primary application stack to a secondary application stack. It's important to make sure that the Route 53 ARC structures that you've added are aligned with the correct resources in your stack, and that everything works as you expect it to. You should test this after you set up Route 53 ARC for your environment, and continue to test periodically, so that your failover environment is prepared, before you experience a failure situation in which you need your secondary system to be up and running quickly to avoid downtime for your users.