You plan a VPC-native GKE cluster that requires IP addresses for both pods and services. What must you provision on the subnet used by the cluster?

For a VPC-native GKE cluster, pod and service IP addresses are allocated from secondary IP ranges defined on the cluster's subnet. You must provision two distinct, non-overlapping secondary IP ranges: one specifically for pods and another for services. This design ensures efficient IP address management and avoids conflicts within the VPC network. A single primary range is insufficient because GKE requires separate IP address pools for pods and services. External IPv6 ranges are not a standard or required component for basic pod addressing in VPC-native GKE clusters, which primarily use IPv4. Cloud NAT instances provide outbound internet access for VMs without public IPs but do not assign internal pod IP addresses.