Dynamic Host Configuration Protocol (DHCP) relay enables you to leverage your existing DHCP infrastructure from within NSX without any interruption to the IP address management in your environment. DHCP messages are relayed from virtual machine(s) to the designated DHCP server(s) in the physical world. This enables IP addresses within NSX to continue to be in synch with IP addresses in other environments.
DHCP configuration is applied on the logical router port and can list several DHCP servers. Requests are sent to all listed servers. While relaying the DHCP request from the client, the relay adds a Gateway IP Address to the request. The external DHCP server uses this gateway address to match a pool and allocate an IP address for the request. The gateway address must belong to a subnet of the NSX port on which the relay is running.
You can specify a different DHCP server for each logical switch and can configure multiple DHCP servers on each logical router to provide support for multiple IP domains.
If the DHCP Offer contains an IP address that doesn't match a logical interface (LIF), the DLR does not relay it back to the VM. The packet is dropped.
When configuring pool and binding at DHCP server, ensure that the subnet mask of the pool/binding for the relayed queries is same as the interface of the DHCP relay. Subnet mask information must be provided in API while DLR is acting as DHCP relay between VMs and Edge providing DHCP service. This subnet mask should match the one configured on gateway interface for VMs on DLR.
DHCP relay does not support overlapping IP address space (option 82).
DHCP Relay and DHCP service cannot run on a port/vNic at the same time. If a relay agent is configured on a port, a DHCP pool cannot be configured on the subnet(s) of this port.