Microservices must be internal to the GKE cluster, allow configuring replica counts, and be addressable uniformly from other microservices regardless of replica count. How should you implement this?

Deploying each microservice as a Deployment allows you to manage replica counts, ensuring scalability and high availability. Exposing it with a Service provides a stable IP address and DNS name, abstracting away individual pod IPs and handling load balancing across replicas. Other microservices can then reliably address it using the Service's DNS name, fulfilling the requirement for uniform addressing regardless of replica count. Using an Ingress is incorrect because Ingress is primarily for external HTTP/S access to services, not for internal cluster communication. Deploying microservices directly as Pods is incorrect because Pods are ephemeral and don't inherently provide replica management, self-healing, or stable addressing like Deployments and Services do.