Get started Bring yourself up to speed with our introductory content.

Build an AWS multiregion architecture to meet global demand


Route traffic with AWS' cloud DNS, Route 53

Source:  Getty Images
Visual Editor: Online Design

Proper traffic management is vital for any clustered application. The focus shifts to routing network traffic between regions when the nodes of an application are distributed, so developers must consider the routing policies that they can apply to their application traffic.

AWS offers a comprehensive domain name system platform through Amazon Route 53, its cloud DNS service that provides access to globally available infrastructure, such as Amazon EC2 instances, Amazon S3 buckets, Elastic Load Balancing services and others. Amazon Route 53 provides DNS failover, domain name registration and DNS health monitoring, and architects can also use it to access a menu of routing policies to orchestrate an application's traffic flow preferences.

The cloud DNS service can route global traffic with up to five different policies, along with simple routing and failover routing. The geolocation routing policy in the cloud DNS service routes traffic based on user location. Ideally, this minimizes application latency by connecting the user to the closest region where the application runs and can help enforce geographic constraints on application and data access. The geoproximity routing policy also routes traffic based on user location, but it can shift traffic to other regions as conditions dictate, too.

The latency routing policy will distribute traffic to regions with the lowest latency. A multivalue answer routing policy would essentially direct traffic to random regions where the application is deployed, but this may mean traffic is sent to a region that is too geographically distant or not compliant with data governance policies. Conversely, a weighted routing policy can route traffic across multiple regions in desired proportions. This can help when regions are not configured with identical resources; for example, a region with more resources may receive a larger proportion of traffic than another region with fewer provisioned resources.

When designing AWS multiregion deployments, application architects must carefully consider these routing policy implications to ensure that traffic is handled in accordance with established business policies and regulatory obligations.

View All Photo Stories