Working with deployment strategies - Amazon AppConfig
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Working with deployment strategies

A deployment strategy enables you to slowly release changes to all targets or specific segments over minutes or hours—either session based or along your own target dimension by leveraging entity-based deployments.

Note

Amazon AppConfig Agent (version 2.0.136060 or later) supports deploying feature flag or free-form configuration data to specific segments or individual users during a gradual rollout. Entity-based gradual deployments ensure that once a user or segment receives a configuration version, they continue to receive that same version throughout the deployment period, regardless of which compute resource serves their requests. For more information, see Using Amazon AppConfig Agent for user- or entity-based gradual deployments.

An Amazon AppConfig deployment strategy defines the following important aspects of a configuration deployment.

Setting Description

Deployment type

Deployment type defines how the configuration deploys or rolls out. Amazon AppConfig supports Linear and Exponential deployment types.

  • Linear: For this type, Amazon AppConfig processes the deployment by increments of the growth factor evenly distributed over the deployment. Here's an example timeline for a 10 hour deployment that uses 20% linear growth:

    Elapsed time Deployment progress

    0 hour

    0%

    2 hour

    20%

    4 hour

    40%

    6 hour

    60%

    8 hour

    80%

    10 hour

    100%

  • Exponential: For this type, Amazon AppConfig processes the deployment exponentially using the following formula: G*(2^N). In this formula, G is the step percentage specified by the user and N is the number of steps until the configuration is deployed to all targets. For example, if you specify a growth factor of 2, then the system rolls out the configuration as follows:

    2*(2^0)
2*(2^1)
2*(2^2)

    Expressed numerically, the deployment rolls out as follows: 2% of the targets, 4% of the targets, 8% of the targets, and continues until the configuration has been deployed to all targets.

Step percentage (growth factor)

This setting specifies the percentage of callers to target during each step of the deployment.

Note

In the SDK and the Amazon AppConfig API Reference, step percentage is called growth factor.

Deployment time

This setting specifies an amount of time during which Amazon AppConfig deploys to hosts. This is not a timeout value. It is a window of time during which the deployment is processed in intervals.

Bake time

This setting specifies the amount of time Amazon AppConfig monitors for Amazon CloudWatch alarms after the configuration has been deployed to 100% of its targets, before considering the deployment to be complete. If an alarm is triggered during this time, Amazon AppConfig rolls back the deployment. You must configure permissions for Amazon AppConfig to roll back based on CloudWatch alarms. For more information, see Configure permissions for automatic rollback.

You can choose a predefined strategy included with Amazon AppConfig or create your own.

Note

Amazon AppConfig Agent (version 2.0.136060 or later) supports deploying feature flag or free-form configuration data to specific segments or individual users during a gradual rollout. Entity-based gradual deployments ensure that once a user or segment receives a configuration version, they continue to receive that same version throughout the deployment period, regardless of which compute resource serves their requests. For more information, see Using Amazon AppConfig Agent for user- or entity-based gradual deployments.

Topics