We already know how to classify Grupo Meridiano's networks: two private client-server LANs (Valencia and Bilbao) joined by a VPN over the public WAN. The question we left open is how the devices connect to one another inside each network: everything onto one shared cable? everything to a central point? each machine to every other? That interconnection layout is called a topology, and it is no cosmetic detail: it determines how much the network costs, how it performs, what happens when a cable or a device fails, and how easy or hard it will be to expand and troubleshoot. In this lesson we will distinguish physical topology from logical topology, walk through the classic topologies (bus, star, ring, mesh, tree and hybrid) with their pros and cons, learn the criteria for choosing among them and, as the module's grand finale, draw the complete topology of Grupo Meridiano on paper.

1. What is a topology? Physical versus logical
2. Bus topology
3. Star topology
4. Ring topology
5. Mesh topology
6. Tree (hierarchical) topology
7. Hybrid topologies
8. Comparison and selection criteria
9. Case study: designing the topology of Grupo Meridiano

A network's topology is the map of how its devices are interconnected. Two planes are worth separating from the very start:

• Physical topology: where the cables (or radio links) actually run; the drawing you would lay over the office floor plan. It answers "what is connected to what?".
• Logical topology: where the data flows, regardless of the cabling. It answers "what path does the data actually take?".

The two need not match. The classic example: in early Ethernet networks built with hubs (dumb repeaters that predate the switch), the cabling was physically a star (every cable ran to the hub), but the hub repeated every piece of data out of all ports at once, so logically everyone shared a single medium: it behaved like a bus. Modern switches brought the two planes into line: the physical star is also a logical star, because the switch delivers each piece of data only to its recipient (as we saw in the first lesson).

Another example we already know: Grupo Meridiano's VPN. Physically, Valencia and Bilbao share no cable at all — each has its own Internet access; logically, the tunnel connects them as if there were a direct link between the two routers. The logical topology adds a link that does not exist on the physical plane.

In what follows we describe the topologies in their physical form, which is how they are always presented, pointing out the logical plane where it adds something.

Every device connects to a single shared backbone cable (the bus), capped at both ends by terminators that absorb the signal.

When a machine transmits, the signal travels the whole cable and everyone receives it; only the intended recipient processes it. And since the medium is shared, if two machines transmit at once a collision occurs and both must resend.

• Advantages: minimal cabling and cost; very simple to set up for a few machines.
• Drawbacks:
  • One fault in the backbone takes down the entire network, and pinpointing the exact break is agony.
  • Collisions skyrocket as the number of machines grows: performance degrades quickly.
  • Expanding means touching the backbone cable, interrupting everyone.

Today the bus is obsolete in office local networks (it was the coaxial-cable Ethernet of the 80s and 90s), but it is still important to understand: it survives in industrial and automotive environments, and its "shared medium with collisions" logic will resurface when we study how wireless networks work, where the air is, for all practical purposes, a bus.

Every device connects to a central node — nowadays, a switch — with a dedicated cable per machine. It is the standard topology of any modern LAN.

• Advantages:
  • A broken cable affects only its own machine: everyone else keeps working. Compared to the bus, that is a massive leap in reliability.
  • Easy troubleshooting: a workstation is down? Check its cable and its port. Switches also signal every port with an LED.
  • Expansion is trivial: one more cable to a free port, without interrupting anyone.
  • With a switch at the center, each machine gets its own bandwidth with no collisions.
• Drawbacks:
  • The central node is a single point of failure: if the switch dies, the whole star goes down.
  • More cabling than a bus: one cable per machine back to the central point (installation cost, trunking, wall sockets).
  • The switch's port count limits growth (though this is solved by chaining switches, as we will see in the tree topology).

It is no accident that the single point of failure is the star's headline drawback: a large part of professional network design consists of deciding which single points of failure you accept and which you eliminate through redundancy, paying the corresponding cost. Hold on to this idea: we will use it in the case study.

Each device connects to the next one, forming a closed circle; data travels from machine to machine, usually in one direction, and each node regenerates the signal and passes it to its neighbor.

• Advantages: no collisions (transmission turns are orderly: in classic Token Ring networks, via a "token" circulating around the ring); predictable, stable performance even under heavy load; each node regenerates the signal, which allows greater distances to be covered.
• Drawbacks: in a simple ring, a single failed node or cable breaks the entire circle; adding or removing machines means opening the ring; troubleshooting is awkward.

As an office topology, the ring (IBM's Token Ring, Ethernet's rival in the 90s) lost the battle to the star. But it is not dead: carrier networks and metropolitan fiber rings use dual rings (two rings running in opposite directions), where if the fiber is cut at one point, traffic loops around the other way in milliseconds. It is a perfect example of how a topology with redundancy turns its weakness into a strength.

Each device connects to several (partial mesh) or to all (full mesh) of the others.

• Advantages: maximum fault tolerance — if one link fails, alternative paths exist; no central bottlenecks; the redundant paths allow traffic to be spread out.
• Drawbacks: extremely expensive to cable and to manage. A full mesh of n nodes needs n·(n−1)/2 links: with Valencia's 20 PCs that would be 190 cables (and 19 network cards per machine). Unworkable as a workstation LAN.

Its natural habitat is elsewhere: where failure is not an option and the nodes are few. The Internet is, at its core, a partial mesh of carrier routers: that is why a fiber cut at one point on the planet doesn't bring the network down — traffic detours along another path. You will also find it in home mesh Wi-Fi and in sensor networks. The rule of thumb: mesh is applied to the critical links between interconnection devices, not to user workstations.

It is a star of stars: several switches, each the center of its own star, connect in turn to a higher-level switch (or router), forming a hierarchy shaped like an inverted tree.

• Advantages: scales in an orderly way (a new floor? a new switch hanging off the main one); it mirrors the building's physical structure (one switch per floor or department); each branch's local traffic doesn't disturb the others; troubleshooting inherits the star's simplicity.
• Drawbacks: the higher up a node sits, the more damage its failure causes (the main switch drags down every branch); the links between levels concentrate the traffic of their entire branch and must be provisioned better (more speed) than workstation links.

It is the topology of any medium or large office building, and the basis of the professional hierarchical design in tiers (access, distribution, core), which we only mention in passing here. For a small site like Valencia, a two-level tree (router → switch → workstations) is more than enough.

A hybrid topology combines two or more of the previous ones in a single network. In practice, almost every real network is hybrid: you use the right topology in each zone according to what is at stake there.

Common examples:

• Star + star joined by a link: two star-wired offices connected to each other (our case, as we are about to see).
• Tree with mesh links: a hierarchy of switches where the upper levels are linked in duplicate to remove single points of failure.
• Ring of sites with a star at each site: typical of organizations with several branches joined by carrier fiber.

The message is that pure topologies are items in a catalog, not religions: the designer mixes and matches.

The table worth keeping at hand:

And the selection criteria, as questions you must ask yourself before any design:

1. What happens if X fails? Walk mentally through every cable and every device. How many users go down? Is that acceptable? Wherever it isn't, redundancy is needed (and must be paid for).
2. How much does it cost? Not just the equipment: the cabling (building work, trunking, certification) is usually the big line item, and it is paid once but shapes the network for 15–20 years.
3. How will it grow? A topology that forces a redesign to add 5 workstations is a bad topology for a company that is hiring.
4. How is it troubleshot? You will build the network once, but you will troubleshoot it hundreds of times. The star and the tree win this contest every time.
5. How much traffic does each link concentrate? Links that aggregate the traffic of many machines (switch to switch, switch to server, site to site) need more capacity than links to a single workstation.

Let's apply the criteria to the inventory from lesson 01-01 and the type map from 01-02. The decisions, with their rationale:

Valencia site (20 workstations, server, printer, AP):

• A star with the switch at the center: each fixed workstation, the server, the printer and the access point, each with a dedicated cable to the switch. It is the standard choice for per-workstation reliability, expansion and troubleshooting.
• The switch–server and switch–router links are aggregation links: the traffic of all 20 workstations passes through them, so they are provisioned at a higher speed than workstation links (criterion 5).
• The access point hangs off the switch like any other machine: the WLAN is, logically, a wireless branch of the same star.
• If Valencia grew to a second floor, the star would evolve into a tree: a second floor switch hanging off the main one. The chosen topology already anticipates that growth (criterion 3).

Bilbao site (5 machines, printer):

• An identical star in miniature: the 5 machines and the printer to the switch, and the switch to the router. Same logic, same ease of troubleshooting, and spare ports for the newly arrived interns.

The link between sites:

• The site-to-site VPN tunnel between the two routers (previous lesson) is, topologically, a logical point-to-point link: on the physical plane there is no Valencia–Bilbao cable, but on the logical plane the company network has a direct link between the two routers. It is the course's best example that physical and logical topology are different planes.

The whole: two stars joined by a point-to-point link = a hybrid topology.

Single points of failure accepted (criterion 1, with small-business honesty): each site's switch, each site's router and each site's Internet connection. For 25 employees, duplicating all of that isn't worth it; the sensible decision is to accept them, know exactly what the impact of each one is, and — as the only cheap improvement — consider a backup Internet line in Valencia, which is where the server lives and, with it, the work of the entire company. This way of reasoning — which failure do I accept, which failure do I pay to avoid — is network design in its purest form.

• Mistaking the drawing for the logical topology: cables arranged in a star don't guarantee that data flows as a star (the old hub was a physical star and a logical bus). Always ask about both planes; with modern switches they match, with Wi-Fi and VPNs they don't.
• Dismissing the "old" topologies as useless: the bus and the ring are no longer cabled in offices, but their logic lives on: Wi-Fi behaves like a bus (shared medium) and carriers use dual fiber rings. Understanding them is understanding those technologies.
• Choosing a topology on installation cost alone: cabling is paid for once; faults and expansions, for decades. A well-cabled, well-documented star pays for itself in the first year of quick diagnoses.
• Ignoring single points of failure... or trying to eliminate them all: both extremes are mistakes. The professional approach is to list them, weigh the impact of each one, and consciously decide which are accepted and which are made redundant.
• Provisioning every link the same: a workstation cable carries one person's traffic; the switch–server link, the whole office's. Aggregation links always call for more capacity.
• Tip: always draw the topology before touching a cable, and keep the drawing up to date. An up-to-date diagram is the cheapest troubleshooting tool there is; we will prove it in module 6.

State which topology (bus, star, ring, mesh, tree or hybrid) describes each scenario, and justify it in one sentence:

1. The 6 Bilbao devices (5 machines + printer), each with its own cable to the switch.
2. A carrier links the exchanges of 5 cities with fiber forming a closed circle with traffic flowing in both directions.
3. A 3-floor building with one switch per floor, all three connected to a main switch next to the router.
4. Four backbone routers of a carrier, each directly linked to the other three.
5. The complete Grupo Meridiano network: the two sites plus their VPN.

Using the final diagram from the case study, state which users lose what in each incident, and rate the impact (low / medium / critical for the business):

1. The cable between a Valencia PC and the Valencia switch breaks.
2. The Bilbao switch fails.
3. The Valencia switch fails.
4. Bilbao's Internet connection goes down.

Grupo Meridiano is considering renting the floor above the Valencia office for 10 new workstations. One vendor proposes: "we run a cable from each new workstation upstairs down to the current switch on the ground floor". Another proposes: "we install a switch on the new floor and join it to the current switch with a single link". State which topology results from each proposal, give one advantage and one drawback of each, and argue which one you would choose by applying the lesson's criteria.

Solution 1:

1. Star: every node with a dedicated link to a central point (the switch).
2. Ring (dual): a closed circle of nodes; the two-way flow provides the redundancy that fixes its weak point.
3. Tree (hierarchical): per-floor stars hanging off a higher level.
4. Full mesh: each node linked to all the others (4 nodes, 6 links).
5. Hybrid: two stars joined by a logical point-to-point link (the VPN).

Solution 2:

1. Only that PC loses the network; the rest of Valencia and Bilbao keep working. Low impact (that is the star's great virtue).
2. Bilbao's 5 machines and printer are cut off from each other and from Valencia and the Internet. Valencia doesn't even notice. Medium impact (a 5-person site at a standstill).
3. The entire Valencia LAN goes down: 20 workstations, printer, Wi-Fi and, above all, the file server becomes unreachable for Bilbao too through the VPN. Critical impact: it is the company's most serious single point of failure.
4. Bilbao keeps its local network (its 5 machines see each other and can print), but loses the Internet and the VPN tunnel — that is, access to the Valencia server. Medium impact: local work possible, work with corporate files impossible.

Solution 3:

• Proposal A (all cables down to the ground-floor switch): the current star, simply bigger. Advantage: a single switch to administer and no new aggregation link that could saturate. Drawbacks: 10 cable runs between floors (expensive building work), it consumes 10 ports on the current switch, and every future expansion repeats the work.
• Proposal B (a switch on the new floor linked to the current one): turns the network into a two-level tree. Advantages: a single cable run between floors and trivial future expansion (free ports upstairs). Drawbacks: a new point of failure appears (the upstairs switch, which would take down its floor), and the inter-switch link aggregates the traffic of the 10 workstations, so it must be provisioned with more capacity than a workstation link.
• Recommendation: proposal B. It applies criteria 2, 3 and 5: less building work, orderly growth (it is the natural star → tree path we already anticipated) and the new switch's risk is contained to one floor. It only requires taking care with the sizing of the inter-switch link.

With this lesson we close the first module, and we now hold the complete blueprint of our network. We know that topology is the way devices are interconnected, and that it must be looked at on two planes: the physical (the cables) and the logical (the flow of data). We have walked the catalog — bus (cheap and fragile, residual today), star (the standard of every modern LAN), ring (alive in carrier networks with dual rings), mesh (maximum fault tolerance, reserved for critical links), tree (the star that scales across floors and departments) and hybrids (what every real network ends up being) — and, above all, we have learned the selection criteria: the impact of each failure, cost, growth, troubleshooting, and the sizing of links that aggregate traffic. With them we have designed Grupo Meridiano: one star in Valencia, another in Bilbao and a logical point-to-point link — the VPN — between the two; and we have honestly listed the single points of failure a small business accepts. But a perfect blueprint is still just a blueprint: we have devices, cables and a topology, and we have not yet explained how the machines understand each other when they finally talk. What "language" do a consultant's PC and the file server speak? Who decides how a conversation starts, how data is sliced up, how an error is detected? Those rules of the game are the communication protocols, and we devote the whole of module 2 to them, starting with the next lesson, Introduction to Communication Protocols.