2.1 Networks Classified by Size

The size of a network drives almost every design choice — hardware, protocols, who manages it, what security threats matter.

Why the distinction matters for security: a LAN is generally trusted internally — your home devices trust each other, and most home routers don't filter traffic between them. A WAN is generally untrusted — you assume anyone on the internet could be hostile. The border between the LAN and the WAN is where firewalls live. Chapter 14 covers that.

2.2 Network Topologies — How Devices Physically Connect

Topology = the shape of a network. Even though two networks might use the same technology, if they're wired in different shapes, they behave very differently — in performance, reliability, and security.

Star topology

Every device connects to a central device (usually a switch). The switch handles all traffic. If any one device fails, the rest of the network keeps working. If the switch fails, everything goes down.

Pros: simple, easy to add/remove devices, one failure doesn't take down the network.
Cons: the central switch is a single point of failure; you need one port per device.

Where you'll see it: Every home Wi-Fi (devices → router), every office (devices → switch → router), every school. This is by far the most common topology today.

Mesh topology

Every device connects to every (or many) other devices. Multiple paths exist between any two nodes, so if one connection fails, traffic reroutes.

Pros: fault-tolerant, no single point of failure, great redundancy.
Cons: expensive, lots of cabling, complex to manage.

Where you'll see it: the internet's backbone (multiple paths between major routers), military/critical networks, modern home Wi-Fi mesh systems. Full mesh is rare; partial mesh (many but not all connections) is more common.

Bus topology

All devices share one cable (the "bus"). Everyone sees every message; each device ignores messages not addressed to them.

Pros: simple, cheap, easy to install.
Cons: the shared cable is a single point of failure. Performance degrades as devices are added. Obsolete for modern wired networks, though Wi-Fi behaves logically like a bus since all devices share the radio medium.

Where you'll see it: legacy ethernet (pre-1990s), still conceptually how a hub operates. Mostly historical.

Ring topology

Each device connects to exactly two neighbours, forming a closed loop. Data travels in one direction around the ring. Popular in early token ring networks.

Pros: predictable timing, no collisions.
Cons: a single cable break breaks the whole ring unless you have a dual ring.

Where you'll see it: some industrial control networks, legacy systems, fibre optic backbones using dual-ring FDDI. Rare in general-purpose networks today.

Tree (hybrid) topology

Multiple star networks connected together by a backbone — a star-of-stars. Each classroom is a star; all classrooms connect up to a main switch; that connects to the router.

Pros: scales well, isolates faults within branches.
Cons: root node failures affect everything below; complex design.

Where you'll see it: Most mid-to-large office networks, schools, campuses. When someone says "star topology" about a large network, they usually mean tree.

Point-to-point

Just two devices connected directly. The simplest possible network.

Where you'll see it: your laptop plugged directly into your modem, a direct WAN link between two sites, a serial console cable.

2.3 How Topology Affects Security

Topology has real security consequences. A network's shape determines:

• How attacks spread. A flat bus-like network lets malware reach every device. A segmented tree network can stop it at branches.
• Where choke points are. Star topologies have an obvious central choke point — perfect for placing a firewall.
• Reliability under attack. Mesh topologies survive DDoS better because traffic can reroute around overloaded links.
• Monitoring visibility. Centralised topologies make traffic easier to monitor; distributed ones are harder.

2.4 Network Architectures — Client-Server vs Peer-to-Peer

Topology is about physical shape. Architecture is about roles — who does what.

Client-Server

Some devices (servers) provide services; others (clients) consume them. The server is always on, stores authoritative data, and enforces rules. Clients connect when they want something.

Pros: Centralised control, easy to manage, authoritative data, easier to secure (one server to protect, with known role).
Cons: Single point of failure (server down = everyone offline), scaling is expensive, the server is a juicy target for attackers.

Peer-to-Peer (P2P)

All devices are equal. Each one can request services from others, and each one can provide services. No central authority.

Pros: No single point of failure, scales naturally (more users = more capacity), hard to censor.
Cons: Inconsistent data, harder to secure (every peer is an attack surface), hard to hold anyone accountable, trust issues.

Hybrid architectures

Most real systems mix both. WhatsApp looks like client-server (you talk to WhatsApp's servers), but messages are end-to-end encrypted so the server can't read them — the security model is peer-to-peer even though the routing is client-server. Modern Skype went the other way: abandoned P2P for client-server, prioritising enterprise features over pure decentralisation.

2.5 Two Other Arrangements To Know

Cloud computing

A form of client-server where the server lives in a massive data centre run by a third party (AWS, Azure, Google Cloud) rather than owned by the organisation using it. Practically, every modern web app is "in the cloud."

Implications: cheaper to start (no upfront hardware), scales on demand, but introduces shared responsibility — the cloud provider secures the platform, you secure your data and configurations. Chapter 16's SaaS startup scenario covers this.

Edge computing

Pushing compute closer to where data is generated, rather than sending everything to a central server. Example: your smart doorbell processes video locally and only sends a summary to the cloud, instead of streaming 24/7.

Implications: lower latency, less bandwidth, but a larger attack surface (many small devices instead of one central one).

2.6 Quick Comparison for Exam Answers

2.7 Quiz Time