2 edited body
source | link

There are different types of NAT. It sounds like you're talking about the standard PAT (port address translation, also known as hide-NATHide NAT) used by private-addressed clients behind an internet-facing firewall.

There will always be (for TCP anyway) a source port and a destination port. The destination port can identify the service ( such as 21 for FTP, 443 for HTTPS, etc.) The source port does not. It is just a random numbered high port.

Source port and IP address are used by remote systems to uniquely identify the connection. With two computers behind a PAT appliance, there's a small chance that both systems might choose the same source port for two different connections to the same remote IP. For this reason, the device must be able to change the source port while keeping track of its association with the original.

There are different types of NAT. It sounds like you're talking about the standard PAT (port address translation, also known as hide-NAT) used by private-addressed clients behind an internet-facing firewall.

There will always be (for TCP anyway) a source port and a destination port. The destination port can identify the service ( such as 21 for FTP, 443 for HTTPS, etc.) The source port does not. It is just a random numbered high port.

Source port and IP address are used by remote systems to uniquely identify the connection. With two computers behind a PAT appliance, there's a small chance that both systems might choose the same source port for two different connections to the same remote IP. For this reason, the device must be able to change the source port while keeping track of its association with the original.

There are different types of NAT. It sounds like you're talking about the standard PAT (port address translation, also known as Hide NAT) used by private-addressed clients behind an internet-facing firewall.

There will always be (for TCP anyway) a source port and a destination port. The destination port can identify the service ( such as 21 for FTP, 443 for HTTPS, etc.) The source port does not. It is just a random numbered high port.

Source port and IP address are used by remote systems to uniquely identify the connection. With two computers behind a PAT appliance, there's a small chance that both systems might choose the same source port for two different connections to the same remote IP. For this reason, the device must be able to change the source port while keeping track of its association with the original.

1
source | link

There are different types of NAT. It sounds like you're talking about the standard PAT (port address translation, also known as hide-NAT) used by private-addressed clients behind an internet-facing firewall.

There will always be (for TCP anyway) a source port and a destination port. The destination port can identify the service ( such as 21 for FTP, 443 for HTTPS, etc.) The source port does not. It is just a random numbered high port.

Source port and IP address are used by remote systems to uniquely identify the connection. With two computers behind a PAT appliance, there's a small chance that both systems might choose the same source port for two different connections to the same remote IP. For this reason, the device must be able to change the source port while keeping track of its association with the original.