You can use endpoint weights with standard routing accelerators in AWS Global Accelerator to control the proportion of traffic directed to each endpoint within an endpoint group. Endpoint weights are numeric values (from 0 to 255) that determine how traffic is distributed among endpoints such as Network Load Balancers, Application Load Balancers, EC2 instances, or Elastic IP addresses. By adjusting weights, you can perform load balancing, conduct canary releases, or gradually shift traffic between endpoints. For example, setting one endpoint's weight to 1 and another's to 255 means the first receives 1/256 of the traffic and the second 255/256. Setting a weight to 0 stops traffic to that endpoint. Global Accelerator calculates the sum of weights in an endpoint group and routes traffic proportionally[1][2][9].
In contrast, custom routing accelerators in AWS Global Accelerator work differently. They route traffic to VPC subnets containing EC2 instances, not to load balancers or other endpoint types. The documentation and API references do not mention the ability to assign weights to endpoints in custom routing accelerators. Instead, custom routing accelerators map incoming traffic on static IP addresses directly to specific EC2 instances within VPC subnets. This implies that endpoint weights are not applicable or supported for custom routing accelerators[6][10].
Summary:
- Standard routing accelerators: Support endpoint weights to control traffic distribution among endpoints in an endpoint group. Weights range from 0 to 255 and affect how much traffic each endpoint receives relative to others.
- Custom routing accelerators: Do not support endpoint weights. Traffic is routed based on mappings to VPC subnets with EC2 instances, without weighted distribution.
Therefore, endpoint weights can be used only with standard routing accelerators, not with custom routing accelerators in AWS Global Accelerator.
Citations:
[1] https://docs.aws.amazon.com/global-accelerator/latest/dg/about-endpoints-endpoint-weights.html
[2] https://docs.aws.amazon.com/global-accelerator/latest/dg/introduction-how-it-works.html
[3] https://www.alibabacloud.com/help/en/ga/user-guide/overview-4/
[4] https://repost.aws/knowledge-center/global-accelerator-failover-different-region
[5] https://boto3.amazonaws.com/v1/documentation/api/1.12.30/reference/services/globalaccelerator.html
[6] https://boto3.amazonaws.com/v1/documentation/api/1.17.39/reference/services/globalaccelerator.html
[7] https://awscli.amazonaws.com/v2/documentation/api/latest/reference/globalaccelerator/add-endpoints.html
[8] https://awscli.amazonaws.com/v2/documentation/api/2.3.2/reference/globalaccelerator/index.html
[9] https://hands-on.cloud/aws-services/global-accelerator/
[10] https://docs.aws.amazon.com/global-accelerator/latest/dg/about-custom-routing-guidelines.html
[11] https://www.alibabacloud.com/help/en/ga/user-guide/create-and-manage-standard-ga-instances