GROUPING ENDPOINTS OF A NETWORK FOR NAT TO ORGANIZE IP ADDRESS SPACE FOR POLICY APPLICATIONS

Techniques and architecture are described that utilize network address translation (NAT) based on a group tag such that legacy and third-party devices may utilize and apply "subnet" based policies, thereby allowing the subnet based policies to be as effective as "group" based policies. In particular, a subnet may be applied to a group tag where the group tag is not understandable outside an access network such as, for example, a fabric network. Thus, when a packet originates from a fabric network utilizing group tags representing source groups of endpoints and is destined for a legacy or a third-party device-based network that does not utilize and/or understand group tags, then the group is converted into a subnet. Since that subnet is different from the source host within the fabric network, network address translation (NAT) is utilized.