Which configuration will meet these requirements?
Create a failover routing policy. Within the policy, configure 80% of the website traffic to be sent to the original resource. Configure the remaining 20% of traffic as the failover record that points to the new resource.
Create a multivalue answer routing policy. Within the policy, create 4 records with the name and IP address of the original resource. Configure 1 record with the name and IP address of the new resource.
Create a latency-based routing policy. Within the policy, configure a record pointing to the original resource with a weight of 80. Configure a record pointing to the new resource with a weight of 20.
Create a weighted routing policy. Within the policy, configure a weight of 80 for the record pointing to the original resource. Configure a weight of 20 for the record pointing to the new resource.
Explanations:
A failover routing policy is used for high availability and failover scenarios, not for controlled percentage traffic distribution. It would direct all 20% of traffic to the failover resource only when the primary resource fails.
A multivalue answer routing policy provides multiple records for a single DNS name and returns multiple values, but it does not allow for controlling the percentage of traffic directed to each resource.
Latency-based routing directs traffic based on latency, not on a percentage distribution. It does not offer a way to control the exact percentage of traffic going to each resource.
A weighted routing policy allows you to distribute traffic between resources based on configured weights. Setting a weight of 80 for the original resource and 20 for the new resource will direct 80% and 20% of the traffic, respectively.
I reckon the answer is:
Create a weighted routing policy. Within the policy, configure a weight of 80 for the record pointing to the original resource. Configure a weight of 20 for the record pointing to the new resource.