Exam AWS Certified Developer Associate topic 1 question 387 discussion

Option B is the best approach to meet the requirements. With this approach, you can update the Amazon Route 53 DNS record entry for the production stage API to use a weighted routing policy. You can set the weight of the first policy to 80, which will route 80% of the traffic to the existing production stage. You can then add a second record for the production domain name and change the routing policy to a weighted routing policy, with a weight of 20. You can then change the alias of the second policy to use the testing stage API, which will route 20% of the traffic to the new production stage API. By doing so, you can minimize the number of errors that any single customer experiences and test the new production stage API before directing more traffic to it. The question never states its intention to move the rest of the traffic over once the testing is complete which would indicate a Canary deployment, so D assumes too much

It's D.

I am inbetween C and D. The question seems like a canary deployment will be appropriate, so D looks good. But I can see that C could also work in this situation.

The typical Canary deployment. First the 20% of trafic then route all of it to the new stage if all went well.

Gateway can handle canary deployment for testing. https://docs.aws.amazon.com/apigateway/latest/developerguide/canary-release.html

C. Deploy an Application Load Balancer (ALB) in front of the REST API. Change the production API Amazon Route 53 record to point traffic to the ALB. Register the production and testing stages as targets of the ALB with weights of 80% and 20%, respectively. This approach would route 20% of the traffic to the new production stage API, and 80% of the traffic to the existing production stage, using the weights configured on the ALB. This would allow for a gradual roll-out of the changes, and minimize the number of errors that any single customer might experience.

D . Canary

I think it's D.