Network design: plan before you buy
Design means deciding what the network must do before buying anything. Good designers ask:
- Who will use it, and how many people now and in 3 years?
- What will they do: email, video calls, a shared printer, a database?
- How fast must it be (bandwidth) and how often can it fail (reliability)?
- How safe must it be, and what is the budget?
Then they choose a layout (topology). A star is easy to fix because each PC has its own cable. A bus is cheap but one break stops all. A mesh is very safe but costs many cables: for n devices a full mesh needs n(n-1)/2 links.
They also make an address plan (which IP range for which room) and a diagram that everyone can read. Always leave room to grow. This is called scalability.
Building the network
Building follows the plan, step by step:
- Install cables (twisted pair, fibre) and sockets, or Wi-Fi access points. Label every cable at both ends.
- Fit devices: switch, router, server, printers. Keep them in a cool, locked rack.
- Configure: give IP addresses, set the router, set passwords, turn on the firewall.
- Test: can each PC reach the server and the Internet? Is the speed what we planned?
- Write down what you did (documentation) and train the users.
Never skip testing. A mistake found now costs minutes; found later it costs a day of lost work.
Operation, monitoring and fault handling
Once users arrive, the work is operation: add new users, take backups, install updates and monitor traffic, speed and uptime with simple tools.
When something breaks, use a calm routine for fault handling:
- Find out what is wrong and who is affected (one PC, one room, everyone?).
- Guess the cause: cable, device, setting, or software. Start with the simplest.
- Fix it, or switch to a spare.
- Test that it is really working.
- Record it so next time is faster.
Useful numbers: availability = (up time ÷ total time) × 100. A network that is up 999 of 1000 hours has 99.9% availability.
System audit: checking the whole system
A system audit is a careful check, by a person who did not build the system, that it works as planned, is safe, and follows rules and laws. The auditor looks at three things:
- Documents: plans, change records, user lists, backup logs.
- Settings: passwords, firewall rules, updates, who can open what.
- Evidence: logs that show what really happened.
The auditor writes a report with problems found and advice. The owner then fixes them and the audit is repeated. Audits are done on a schedule, so the network keeps improving.
Try it: draw your home network
On paper, draw your home network. Put the Wi-Fi router in the middle. Add a dot for every phone, TV and laptop. Count the devices. Now ask: if two more people move in, will it still be fast? Then switch the router off for one minute and note who is affected. You just did design, a fault test and a small audit.
Key formulas and definitions
- Full mesh links = n(n-1)/2
- Star: cables = n (one per device)
- Availability % = (up time ÷ total time) × 100
- Transfer time (s) = file size (bits) ÷ speed (bits per second)
- Usable hosts in a /24 network = 256 − 2 = 254
Worked examples
1. An office has 12 PCs in a star layout. How many cables go to the switch?
In a star each PC has its own cable. So 12 cables.
2. A designer wants every one of 6 devices linked directly to every other. How many links?
n(n-1)/2 = 6 × 5 / 2 = 15 links. This is why full mesh is costly.
3. A server was up for 998 hours out of 1000 hours. Find its availability.
998 / 1000 × 100 = 99.8%.
4. A 100 MB file is sent over a 100 Mbit/s link. How long does it take?
100 MB = 100 × 8 = 800 Mbit. Time = 800 / 100 = 8 s (in ideal conditions).
5. In an office, one PC cannot reach the server but others can. List what to check first.
Only one PC is affected, so look near that PC: 1) its cable and port light, 2) its IP setting, 3) its network card. The server and switch are fine because others work.
6. An audit has 40 checks. 34 pass. What is the pass rate and what should happen next?
34 / 40 × 100 = 85%. The 6 failed checks go in the report, the owner fixes them and the auditor checks them again.
Common mistakes
- Buying devices first and planning later. Plan first, so you buy only what you need.
- Planning only for today. Leave room for more users (scalability).
- Skipping tests and documents after building. Nobody can fix what nobody understands.
- Thinking an audit is only for finding blame. Its job is to make the system better and safer.