FTI’s original website carried a line on its homepage that sums up how our founder, Scott Henderson, has always approached this work: “Networks work because they are properly designed and deployed, not because a vendor’s sales person said they would.”
It has been our working principle ever since, and it has aged well. The equipment has changed beyond recognition. The way networks fail has not. They fail because nobody sat down and worked out what the network had to do before someone bought the hardware to do it.
A quote is not a design
Here is a familiar pattern. A business has slow Wi-Fi, or a new office, or an old firewall. Someone asks a vendor for a quote. The quote arrives with a list of boxes: a firewall, a couple of switches, some wireless access points, and a support contract. It looks professional. It is priced. It gets approved.
What the quote usually doesn’t contain is any evidence that someone asked how many people will use the network, what they do on it, where the walls and the dead spots are, how much Internet bandwidth the business actually has, or what happens when one piece fails.
That isn’t necessarily anyone’s bad faith. A quote is built to sell products, and the person preparing it is often rewarded for selling products. There are good, careful vendors and good salespeople. But a list of equipment answers the question “what can we sell you?” A design answers a different question: “what does this business need, and how do we know this will deliver it?”
What network design actually involves
Design sounds grand. In practice it is a series of plain questions, answered with care and written down.
Requirements first
Who uses the network, from where, and for what? An office running a hosted phone system, video calls and a cloud accounting package has very different needs from one that mostly sends email. Guests, cameras, card readers and printers all count. So do staff who work from home or from another office.
Capacity planning
Capacity planning means working out how much traffic the network must carry, at its busiest, with room to grow. That covers the Internet connection, the links between switches, and the wireless coverage in every room where people actually sit. In much of Alaska, bandwidth is expensive, so the design should spend it wisely rather than assume more can always be bought.
Segmentation and security
A well-designed network separates things that shouldn’t talk to each other. Guest Wi-Fi stays away from the accounting PC. Cameras and other connected devices sit on their own segment (a separate section of the network). The firewall is configured for the business, not left on its factory settings. Security tools reduce risk; good structure makes them work better.
Failure planning
Every component fails eventually. What happens when the Internet connection drops, a switch dies, or the power goes out in February? A design decides in advance which failures the business can live with for an hour and which it can’t, and spends money accordingly.
Testing before buying
Where it matters, an engineer tests before committing: measuring the existing connection, surveying wireless coverage, checking that new equipment works with the systems already in place. It is far cheaper to find a problem on paper, or on a test bench, than after the purchase order is signed and the installer has gone.
Deployment and documentation
The second half of Scott’s line matters as much as the first: properly deployed. Good equipment, badly configured, is still a bad network. Deployment means careful configuration, testing once it is installed, and documentation that records how everything is set up, so the next person who touches it can understand it.
Equipment comes last
This is how we work at FTI. We start by designing the network, including drawings of how it will be laid out and connected. We don’t buy a pile of boxes and start plugging in wires to see what happens. The decision about which equipment to buy is often the last thing we address, because only once the design is settled do we know what the equipment has to do.
Signs your network was sold, not designed
- Wi-Fi is fine in some rooms and useless in others, and the answer was “add another access point”.
- Nobody can show you a diagram of how the network is laid out.
- Guests, staff and devices all share one network.
- The firewall is doing far less than it could, because nobody configured the features you paid for.
- Every problem is met with a recommendation to buy something newer.
- When one device fails, everything stops, and nobody had planned for that.
Questions to ask before you approve a network quote
- What requirements is this based on? Who gathered them, and how?
- How did you size it? What peak usage and growth did you assume?
- What did you measure or test before choosing this equipment?
- How will guests, staff and devices be separated?
- What happens when the Internet connection, a switch or the power fails?
- Who configures it, who tests it after installation, and who documents it?
- Will I receive the documentation, and does it belong to me?
If the answers are mostly about product features and discounts, you are looking at a sales proposal, not a design. Our guide on how to choose a managed IT provider has more questions along the same lines.
How FTI does it
FTI started in 1990 as an engineering software company, and we still approach networks the way engineers approach any system: requirements, design, testing, deployment, documentation. Our engineers design the network before anyone talks about equipment, secure it with Sophos firewalls and Wi-Fi, and keep watch on it through our Kaseya-based monitoring as part of our managed IT services. When the right answer is to reconfigure what you already have, we say so. You can read more about who we are on our about page.
