Posted On: May 19, 2020

You can now select the Traffic splitting deployment policy when deploying application code using AWS Elastic Beanstalk. This is in addition to the four existing deployment policies that Elastic Beanstalk currently supports: All at once, Rolling, Rolling with additional batch, and Immutable.

You can now use the Traffic splitting deployment policy to update an application running in an Elastic Beanstalk web server environment that uses an Application Load Balancer. This deployment policy is well suited for updates in production environments where you want to reduce the risk of application failure when committing changes to your application code. A traffic-splitting deployment allows you to monitor the health of your new application version under a configurable percentage of production traffic before completing the deployment. In case of deployment failures, a traffic-splitting deployment triggers an automatic rollback mechanism. There’s never any application downtime; there are instances serving incoming traffic throughout the deployment process, with a changing load on the old and new application versions.

When using a traffic-splitting deployment, you can specify the percentage of production traffic and the amount of time to monitor a new application version before completing the deployment. Once the deployment starts, you can monitor the health of your new application version by tracking Elastic Beanstalk events in real time. You can also obtain the traffic-splitting deployment events from Elastic Beanstalk logs. To get started, choose the traffic splitting deployment policy using the Elastic Beanstalk console, Elastic Beanstalk CLI, or through the Elastic Beanstalk API. Learn more about the traffic splitting deployment policy in our documentation. See Release Notes for additional details. 

You can follow the public AWS Elastic Beanstalk roadmap on GitHub to get updates on recently launched features, upcoming features, and to provide feedback on what you want us to support. More information on AWS Elastic Beanstalk at: