What is meant by Access Network and Edge Router?

Question

I am reading the book 'Computer Networking A Top Down Approach 8th Global Edition' and the book refers in the Chapter 1.2.1 to 'Access Networks' informally as "the network that physically connects an end system to the first router (also known as the 'edge router') on a path from the end system to any other distant end system.".

I can't seem to exactly grasp what the book means with 'access networks' or 'edge routers' because it proceeds to classify DSL, HFC and FTTH as access network technologies since they provide a connection to the ISP (so the edge router is on the ISP?) but it also classifies WiFi as an access connection (so the edge router is in the home of the user?) even though WiFi does not provides on its own a connection to the ISP.

I already surfed the internet and also asked ChatGPT a series of questions to try to see what exactly is meant by access networks and edge routers. But they all provide very vague definitions. Is a home router considered an edge router, or is an edge router some special router located at some ISP infrastructure. Or are both of these considered edge routers?

So my question is: What exactly (technically) is considered an access network and what exactly is considered an edge router?

Answer 1

As the book states, an access network connects end nodes - nodes with a single network interface. That is in contrast to an interconnecting network (possibly core or distribution) that connects intermediate nodes with multiple interfaces.

End nodes originate (source) or consume (sink) data, intermediate nodes forward data elsewhere.

An edge router is a router between an access network (home, small business, ...) and an intermediate network (large business, ISP). Most often, edge router refers to the ISP's/business's edge, so its the last router in front of the home/SMB routers.

Answer 2

The easiest way to conceptualize access networks vs. edge networks is to think of the direction of traffic flow.

It's easiest to start from the top down:

• An "edge network" generally refers to a physical and/or logical network segment that is Internet/outbound-facing.
• The term almost defines itself; it's any network segment facing your "outside" (edge) interface(s)

On the other hand...

• An access network is typically internal/LAN-facing. An access network segment connects any number of end nodes to the network, which may be workstations, printers, etc.
• Unlike an edge network, an access network segment is any segment facing, or consisting of, the "inside" interfaces
• An access network would be what end users connect to

In the three-tier and collapsed core architectures, the edge network segments would exist at the Core layer. The access network segments would be located at the Access layer.

In the case of your example, in which DSL, HFC, FTTH, and Wi-Fi are described as access network technologies, you are correct. They are labeled such because they provide access to an ISP.

In this particular example, yes, I'd say the router(s) and other infrastructure at the ISP would be the "edge router" in this case. There are types of Wi-Fi that can provide an uplink; a MAN, or Metropolitan Area Network, is one such example.

Be aware that different use cases may imply a different context and/or meaning. These are general terms.

Answer 3

Think about a city with millions of homes and small businesses connected to the Internet. Each one has an individual home router. The ISP has routing equipment in a relatively small number of data centers, where it connects to hosting providers, other ISPs, long-distance links, etc.

The data center does not have space for half a million incoming cables. The ISP does not have a router with half a million separate ports on it. And besides even if you spent the money to put half a million cables underground, they wouldn't fit. They don't fit on telephone poles, either. As you can see, just getting the connections from the individual homes to a data center is a big challenge in itself. The part of the network that solves this challenge is called the access network. Because it needs to host a large number of connections as cheaply as possible, it has different design considerations from the "real" network.

There are many ways it can be built, but for example, in one place I lived, each house had a phone line connection to a roadside cabinet (roughly one per street). The cabinets contained equipment to terminate POTS (phone service) and DSL, and then fiber-optic connections to each other and eventually to a data center, in a shallow tree configuration rooted at the DC. DSL was (and is) used because it doesn't require a new type of cable to be installed to every house. Fiber cable is still installed in the rest of the tree, because there is much less cable needed.

Access networks are typically relatively dumb - effectively giving each house a separate VLAN. The smarts are contained in the data center equipment where it is easier to service. If I send a packet to my neighbor, it typically gets tagged with my connection number somehow and then travels up the tree to the ISP's router at the DC, which recognizes it's on a connection on my tree, somehow tags it with that connection's number and sends it back down the tree.

In a corporate environment, the access network typically isn't as dumb. You may have a proper Ethernet switch or even a router in each office area, so packets can "short-circuit" if they don't need to go all the way to the root. Still, their primary purpose is to connect all the hundreds of computers back to the one or two core switches in the server room.