NAT stands for Network Address Translation. It is the translation of IP addresses and TCP/UDP ports in the headers of IP packets.

Source NAT is the translation of source IP addresses and TCP/UDP ports in the headers of IP packets.

Destination NAT is the translation of destination IP addresses and TCP/UDP ports in the headers of IP packets.

Static NAT is the symmetrical translation of flows for a given IP prefix whether it is initiating the flows as the source, or receiving them as the destination.

PAT stands for Port Address Translation. It is a subset of NAT where TCP/UDP ports are translated. The most typical variant of PAT is source PAT where a larger private source address range is overloaded to a smaller public address range. This is typically accomplished through PAT overloading. PAT overloading occurs when more than one private IP address shares the same public IP address. This is achieved by using different TCP/UDP ports of the same public IP address.

A pool is an IP prefix, IP range, or collection of IP prefixes and/or IP ranges that can be used for translating either the source or destination IP addresses in the headers of IP packets. Pools may also define TCP/UDP port ranges for PAT.

Interface NAT is accomplished through the source NAT feature set in Junos for the SRX.

Source NAT is best used for overcoming IPv4 address exhaustion by hiding the source IP addresses of privately addressed hosts behind one or more public IP addresses.

Destination NAT is the most effective tool for hiding the real IP addresses of systems providing services for other devices on the network.

Static NAT is the most effective tool for integrating overlapping private IP address ranges. Static NAT can most easily be applied to large IP prefixes where hosts will need to both initiate and receive translated traffic.