GUIDE · CORPORATE NETWORKING

Designing the corporate network: topology, segmentation and redundancy

After this guide you will know which decisions to make before buying equipment, how to divide the network into separate parts, where redundancy is needed and in which order to tackle the project. It also points to the topics that go deeper into each part, to be read in the right sequence.

Updated october 2026
10 min read
Altovar Technical Editorial Team

1. Why network design comes before the equipment

The network is the system that connects everything else: workstations, printers, phones, servers, cameras, internet access and often cloud services. When it is not designed, it grows by accumulation: a switch added in a room, an access point placed where a socket happened to be, an improvised link for a new department. After a few years nobody knows any more where a cable runs or why a service works only from one workstation. The design exists to prevent this: decide first how the network should be built, then choose the equipment that realises it.

The design does not require a full-time network engineer. It requires a written description of what the network must do today and in two or three years: how many users, which sites, which critical applications, which connected devices, what type of internet connection. With this document you can compare suppliers' proposals on equal terms, rather than on a list of equipment chosen by whoever is selling it.

2. Requirements and inventory: what the network must do

Before drawing anything, list what is already connected. An inventory of existing devices and connections often shows more than expected: forgotten equipment, test links that became permanent, personal devices on the corporate network. Do it with the help of whoever currently manages the network and with a physical walk through the cabinets, because existing documentation is often out of date. This list is the starting point for understanding which traffic flows are real and which are hypothetical.

Then define the requirements in concrete terms. For each service write who uses it, where it is located, how available it must be and what happens if it stops for an hour. A service that blocks production has different requirements from a department printer. No precise percentage is needed: a shared scale is enough, for example critical, important or ancillary, to be agreed between those who use the service and those who manage it. These answers guide the later choices, especially redundancy and segmentation.

3. Topology: how to connect sites, cabinets and equipment

Topology describes how the equipment is connected to one another. In corporate networks the most common model is hierarchical: a central core, distribution switches per floor or area and access switches to which workstations connect. The model is simple to understand, easy to document and to grow. It does require that every network cabinet has a link to the core and that no point depends on a single undocumented cable.

For multiple sites the choice depends on the links between them. A main site with a few branch offices may use dedicated circuits or encrypted connections over the internet. Ask the provider how it routes traffic between sites, where the exit to the internet is and what happens if that link fails. There is no topology valid for everyone: the criterion is which traffic must pass where and what delay is acceptable for each service. These questions quickly show whether the project is designed or merely drawn.

4. Segmentation: separating what must not talk to each other

Segmentation divides the network into separate parts, usually through VLANs (virtual local area networks, a logical separation on the same equipment). The most common separations are: office workstations, guests, appliances, system devices such as cameras or access control, and servers. The aim is to limit what a compromised device or a visitor connected to the network can reach. The boundaries between segments should be drawn around functions, not around the order of cables or the position of rooms.

Segmentation works only with written rules between segments. Creating VLANs is not enough: you must define which traffic is allowed between them, for example from a workstation to the print server but not the other way round. Rules must be documented and reviewed when a service changes, otherwise they become permanent exceptions. A rule left open for convenience and forgotten after a test is one of the most frequent causes of internal exposure.

5. Redundancy: where it is needed and where it is not

Redundancy means having a second path or a second device for a point that cannot be allowed to stop. Not everything should be duplicated: duplicating everything costs money and complicates management. It is usually best to start with the single points that block many workstations, such as the internet connection, the network core and the power supply to the main cabinets. The criterion is simple: if a failure stops many people or a critical service, that point should be designed with an alternative.

Badly configured redundancy can be worse than no redundancy, because two paths that do not know how to coordinate can create loops or blockages that are hard to diagnose. Ask how the system behaves when a device fails, not just how many devices there are. A controlled failure test, carried out in a maintenance window, tells you far more than any datasheet. Note how long it takes for traffic to flow again and which services remain slow in the meantime.

Critical pointTypical alternativeTo verify with the provider
Internet connectionSecond connection with a different operatorWhat happens at the switchover and who manages the change
Network coreTwo devices configured to work togetherHow the system behaves when one of the two fails
Cabinet power supplyUninterruptible power supply on the main cabinetReal autonomy and periodic battery testing
Internal servers and servicesTwo systems or a copy in another roomRecovery time stated in writing

6. Power, cabling and environment

A network designed on paper often fails for physical reasons: cabinets without ventilation, power without protection, cables laid without labels, switches plugged into a shared socket with high-consumption appliances. The design must record where each cabinet is, what power it has and how it is protected. Structured cabling, meaning the fixed cable system running through the building, must be tested after installation, because a faulty cable cannot be seen from outside.

Devices powered through the network cable, such as phones, access points and cameras, increase the load on the switches. Check the available power per port and the total budget of the switch before buying it, not afterwards. Equipment chosen without this check may power only part of the planned devices, and the fault appears when someone plugs in the last telephone. For details there is a dedicated deep dive on power over the network.

7. The most common mistakes in network design

The errors that recur most are not technical in the strict sense. They almost always stem from choices made without a written document and without considering who will manage the network over the following two years. The design that looks complete on paper often lacks three things: an owner for each part, a list of rules between segments and a procedure for failures. Recognising these errors in advance costs far less than correcting them afterwards.

  • Designing around the equipment available instead of the requirements of the services.
  • Creating segments with no rules between them, leaving everything reachable from everywhere.
  • Adding redundancy everywhere without verifying what happens on failure.
  • Not documenting addresses, VLANs and links, so nobody knows what to touch any more.
  • Leaving equipment administration reachable from the user network.
  • Choosing the lowest price without comparing support, updates and warranty.

8. How to verify that the design works

Verification is not meant to show that the network is perfect, but to demonstrate that it does what the requirements document says. Start with the critical services: for each, verify that it is reachable from the intended segments and not from those that should not see it. Repeat the tests after every significant change, not only at handover. A partial result should be recorded as such, with a list of the tests not yet carried out, rather than presented as a complete acceptance.

Every test must leave a trace: date, outcome, who carried it out and what was corrected. This trace is the basis of the network documentation and becomes indispensable when a new technician needs to understand how the network works or when the supplier changes. Without a trace, today's verification is of no use tomorrow. Keep it with the other technical documents and update it together with the inventory.

  1. Verify that each critical service is reachable from the intended segment and not from the others.
  2. Run a failure test on every redundant link and record the recovery time.
  3. Check that equipment administration is possible only from the management network.
  4. Compare the real inventory with the documentation and correct the differences.
  5. Have the result reviewed by someone who did not take part in the project.

9. How to start

The first step this week is the inventory: list cabinets, equipment, internet connections and the services that cause the most damage if they stop. Write them in a shared document with the IT manager and with the people who use those services. You do not need a complete project to begin: you need to know where things stand now. The manager should be able to say, on one page, which services are covered by the project and which remain outside its scope.

After the inventory, read the topics that go deeper into this guide, in the order you need them. If the network includes Wi-Fi, start with the radio survey. If you are about to lay cables or test existing ones, read the guide on cabling certification. Then tackle power over the network, quality of service, the move to IPv6, the choice between managed and unmanaged switches, monitoring, documentation and sizing of the internet bandwidth.

Frequently Asked Questions

How long does a network design project take for an office?

It depends on size, the number of sites and how well documented the existing network is. For a small organisation the main work is the inventory and defining requirements, whereas a larger network needs physical surveys and acceptance tests. Ask the provider for a phased plan, stating what is delivered at each phase, rather than a single estimate.

Should I design the network before buying equipment?

Yes, in most cases. Buying first means choosing equipment without knowing which services it must support, what power is needed or how to separate traffic. The design can be brief, but it must exist and stay within the scope of the requirements. Purchases should follow from the design, not the other way round.

How many network segments do I need?

There is no right number. The criterion is to separate what has different levels of risk or need: users, guests, appliances, physical security systems and servers. Each additional segment increases management complexity, so it must be justified by a precise rule. A few well-managed segments are better than many with no rules.

Does every small company need redundancy?

Not necessarily. It should be assessed according to what an interruption of the internet connection or of the network core costs for daily work. In many cases the most useful choice is a secondary connection for internet access only and an uninterruptible power supply on the main cabinets. Compare the cost with the damage of an outage, not with the price of the equipment.

Who should own the network design?

An internal manager, even part-time, who approves changes and keeps the documentation up to date. The provider carries out the technical work and must deliver diagrams, configurations and procedures. Management decides the priorities when requirements conflict. If nobody holds this role, the design becomes outdated within a few months.

Can I do everything myself with an internal technician?

You can do the inventory, the requirements and the documentation. For the complete design, cable verification and failure testing, consulting an external provider helps avoid blind spots. The internal technician is often the most useful person precisely because they know the company, but they should not be the only one checking their own work.

About this content

General informational guide. Regulations and technical requirements change over time: always check the official sources and assess your specific case with a qualified technical or legal advisor.

GET IN TOUCH

Talk to our team.

We'll show how Altovar integrates into your infrastructure. No commitment, no sales pitch. Just a technical conversation.

BOOK A DEMO →