performs a peering between the two VPC's (the statement makes sure that this option is feasible since it clearly specifies that there is no overlapping between the ip ranges of both vpc's), deploy the LoadBalancer as internal with the annotation, and configure the endpoint so that the compute engine instance can access the application internally, that is, without the need to have a public ip at any time and therefore, without the need to go outside the google network. The traffic, therefore, never crosses the public internet.
https://medium.com/pablo-perez/k8s-externaltrafficpolicy-local-or-cluster-40b259a19404
https://cloud.google.com/kubernetes-engine/docs/how-to/internal-load-balancing clients in a VPC network connected to the LoadBalancer network using VPC Network Peering can also access the Service
https://cloud.google.com/kubernetes-engine/docs/how-to/service-parameters