Create a TargetEndpoint with a weighted load balancing algorithm. Configure the API proxy to use the same weights for each region's backend.

Configure a regional internal Application Load Balancer in each region, and use health checks to verify that each backend is active. Create a DNS A record that contains the IP addresses of both regions' load balancers. Configure a Targetserver for each region that uses this DNS name.

Configure a global external Application Load Balancer and configure each region’s backend with a different regional backend service. Each region communicates to this single global external Application Load Balancer as its TargetServer.