Highly available AWS services are supposed to be build covering multiple Availability Zones (AZ). Highly available AWS services with single point of failures (example: traditional, non parallel database) need to implement their single point of failure in each Availability and assure that the state of the services stays syncronized.

The second task is to assure that the backup services becomes available to end consumers in the case of a failure of the primary service.

One way to solve this problem is to reassign a given IP adress to the standby server. AWS currently offers two ways to accomplish this failover. The IP Failover scenarios are described here.

This document focuses on a name based failover in a private VPC with access to an on premises intranet.

Targeted architecures for Route53 Failover

This document describes how to switch the IP address for a given name entry in the AWS specific Domain Naming service Route53. This approach will work for the following architectures:

• Consumers of a highly available service connect through the services through name exclusively.
• Consumers of such a service have a comain name service configured which uses the AWS Route52 information
  • They use a public AWS IP address with a publicly registered domain
  • They have an on premises intranet which uses AWS Route53 as a secondary domain name server.
• End consumers are able to reconnect a network session in a case of failure

The solution presented on the following pages will resonate more with DNS users. A similar solution using Active Directory and the Actice Directory Connector is being presented here.