Network Theory – Broadcast Link vs. Point-to-Point Links

ethernetNetworknetwork-theorysubnet

What is the distinction between broadcast links and point-to-point links? In the paragraph below, the author switches between "broadcast link" and "broadcast network" which is causing my confusion..is there a distinction between a "broadcast network" and a "broadcast link"? The word 'link' to me implies a physical link.

Does a broadcast network consist of point-to-point links?

Can a network be made of up of point-to-point links and broadcast links?

Broadly speaking, there are two types of transmission technology that are in widespread use: broadcast links and point-to-point links.
Point-to-point links connect individual pairs of machines. To go from the source to the destination on a network made up of point-to-point links, short mes- sages, called packets in certain contexts, may have to first visit one or more inter- mediate machines. Often multiple routes, of different lengths, are possible, so finding good ones is important in point-to-point networks. Point-to-point transmission with exactly one sender and exactly one receiver is sometimes called unicasting.

In contrast, on a broadcast network, the communication channel is shared by all the machines on the network; packets sent by any machine are received by all the others.

— From, Computer Networks – A. Tanenbaum

Best Answer

Tanenbaum is referring to a rather theoretical and somewhat narrow view of a network. In that, each connection (link) can be either point-to-point (p2p) or broadcast (p2mp).

A p2p link can only connect two nodes - think of a simple serial line. A packet sent by a node is only received by the node at the other end. Addressing is an option but not necessarily required.

With a broadcast link, more than two nodes can be connected - either by all sharing a single wire (shared medium) or with some kind of concentrator (hub) in between. This is also called point-to-multipoint (p2mp). That way, a packet sent by a node is received by all other nodes. Obviously, there must be some kind of addressing but that may happen at a higher network layer.

In real life, it helps to look at a specific technology. The ubiquitous Ethernet started out as a broadcast network in the physical layer (10BASE5 and 10BASE2) - a shared electric wire is tapped into by all nodes (bus topology), and a packet sent by any node was received by all the other nodes. Each packet has a destination address and packets not addressed to the receiver were simply ignored.

Because shared media can become a pain when troubleshooting and to leverage existing twisted-pair cabling, that shared wire was replaced by a repeater hub and all taps were replaced by electric p2p links (10BASE-T). The hub took care that on the handling side (upper physical layer) nothing changed but the electric topology (lower physical layer) changed from bus to star.

A simple broadcast network ("collision domain") can only transport a single packet at any given time, regardless of its size - that doesn't scale well. Larger networks were split by bridges or consequently by switches. Those split up the shared collision domain into one on each port and selectively forward between them by destination address. That way, many packets (frames) can run through a network simultaneously. Here, each link represents a p2p connection at the full physical layer (L1). However, addressing is still used throughout, so at the data link layer (L2) the network is still a kind of broadcast network (a selective one though).

Does a broadcast network consist of point-to-point links?

It may but not necessarily though (e.g. 10BASE5 or Wi-Fi).

Can a network be made of up of point-to-point links and broadcast links?

Yes - at least with "network" as a higher-layer term. An IP network (L3) can use a wide variety of links further down. Even an Ethernet network can use a mix of p2p and p2mp L1 links (e.g. with 10BASE-T and 10BASE5 connected together).