What Is Networking?
What Is Networking?
A network moves data between endpoints over links, using protocols—agreed rules for format, addressing, and behavior. Everything later in this book (VLANs, OSPF, VXLAN) is a specialization of that idea.
Learning goals
- Name the main building blocks of a network and what problem each solves
- Distinguish local delivery (same L2 domain) from routed delivery (across networks)
- Explain why we use layered protocols instead of one monolithic format
- Tie lab tools (Containerlab,
ip, captures) to those ideas
Building blocks
| Piece | Role | Lab example |
|---|---|---|
| Endpoint / host | Source or sink of traffic | Alpine container, your laptop |
| Interface | Attachment to a link | eth0, eth1 |
| Link | Path between interfaces | Cable, veth pair, Wi‑Fi |
| Switch / bridge | Forward within an L2 domain by MAC | Linux bridge, virtual switch |
| Router | Forward between L3 networks by IP | FRR node, Linux with forwarding |
| Protocol | Rules and message formats | Ethernet, IPv4, TCP, DNS |
What problem networking solves
Without a network, processes on different machines cannot cooperate. Networking provides:
- Reachability — a path exists (or is reported as unreachable)
- Naming / addressing — who is who (MAC, IP, names via DNS)
- Multiplexing — many conversations share links (ports, VLANs, queues)
- Reliability options — best-effort (UDP/IP) vs recovered streams (TCP)
- Policy — who may talk to whom, with what priority
Switching vs routing (first cut)
Same subnet / VLAN Different subnets
───────────────── ─────────────────
Host A ──switch── Host B Host A → gateway → … → Host B
Frame uses MAC Packet uses IP; each hop rewrites L2
- Switching (L2): deliver a frame inside one broadcast domain; learn MACs; flood unknown unicasts.
- Routing (L3): deliver a packet across networks; each router decrements TTL and chooses a next hop.
You will refine both in the Layer 2 and Layer 3 parts. Foundations give vocabulary and host-level truth.
Protocols as contracts
A protocol answers:
- What bits mean (headers, fields)
- Who may send what next (state machines)
- How errors are signaled (ICMP, TCP RST, DHCP NAK)
Vendors invent CLIs; RFCs and open implementations define the contracts this book cares about.
Why labs beat slideware
Networks fail in partial, asymmetric, and time-dependent ways. A Containerlab topology lets you:
- Change one variable (shut a link, wrong mask, bad gateway)
- Observe tables (ARP, routes, MAC) and packets (tcpdump)
- Reset and try again
Observation toolkit (preview)
| Tool | Use |
|---|---|
ip addr / ip route / ip neigh |
Host addressing and neighbors |
ping / traceroute / mtr |
Reachability and path hints |
ss / netstat |
Sockets and listeners |
tcpdump / tshark |
On-the-wire proof |
| Containerlab | Multi-node topologies as code |
Checkpoint
Before deeper chapters, you should be able to point at a simple two-host lab and say: these are endpoints, this is a link, delivery is L2 or L3 for this destination, and here is how I would prove it.