Port forwards

Port forwards allow devices on the public side of NAT (NAT outside) to initiate sessions toward those devices, usually servers, that are hidden behind NAT (NAT inside). Another term for port forwards is NAT pinhole.

A port forward represents a previously created (before any traffic is received from the inside) mapping between a TCP/UDP port on the outside IP address and a TCP/UDP port on the inside IP address assigned to a device behind the NAT. This mapping can be created statically by configuration (such as CLI, MIB, YANG, or NETCONF), or it can be created dynamically with protocols such as PCP or UPnP. Port forwards are supported only in NAT pools in Network Address and Port Translation (NAPT) mode. NAT pools in 1:1 mode do not support configured port forwards because, by default, the pools allow traffic from the outside to the inside and this cannot be disabled. Pools in 1:1 mode (whether protocol agnostic) do not perform port translation; therefore the inside and outside always match.

In UPnP, the forwarded ports are created with the port range of the NAT subscriber, whereas with PCP and Static Port Forwards (SPF), the forwarded ports are allocated from a dedicated port range outside of the port blocks allocated to individual NAT subscribers. There are two ranges dedicated to port forwards in NAT:

Port forwarding ranges (well-known ports and wildcard ports) are shared by all NAT subscribers on a specific outside IP address. Port blocks that are individually assigned to the subscriber cannot be allocated from the port forwarding range. The wildcard port forwarding range can be configured only when the pool is administratively disabled.

See the Port Control Protocol (PCP) and Universal plug and play Internet Gateway Device service sections as well as the SR OS R22.x.Rx Software Release Notes for more information about these protocols and the supported NAT types.