How an intranet can be invisible to the IP module if the IP module allows a non-networked device connect to a network system?

Question

A quote from RFC 791 document "Fragmentation, transmission and reassembly across a local network which is invisible to the internet protocol module is called intranet fragmentation and may be used."

A quote from easytechjunkie site "An Internet protocol (IP) module is a component that allows a non-networked device to connect to a network system."

How a local network, called intranet can be invisible to the Internet Protocol module if the Internet Protocol module allows a non-networked device connect to a network system?

As I see, a local network cannot be created if IP module can't see the intranet. But I don't think that RFC Document can contain false information. So I confused and don't know how to understand this strange situation.

Answer 1

Re RFC 791: it's the fragmentation, transmission and reassembly that may be invisible to a node's IP stack, not the local network itself.

That internet protocol module is usually called IP stack today. It's the functionality that enables connectivity to a local network. Back in 1981 it was entirely external to the system being networked. For a long time now (roughly 15 years later), it's become an integral part of most operating systems.

Answer 2

To expand on @Zac67... RFC791 is pointing out fragmentation can happen at layer-2 or layer-1. Such fragmentation (and reassembly) would be invisible to any higher layer protocols. This is an important distinction as the IP "DF" (do-not-fragment) flag does not apply to other layers.

Such fragmentation is rare. For example, MLPPP (multi-link PPP) can fragment frames for parallel transmission. ATM can be seen as fragmentation, as everything has to be divide to fit ATM cells. If you consider IPSec or any tunneling protocols, they can fragment traffic as well -- fragmenting at the outer layer is invisible to the inner layer / payload.