A network infrastructure that no longer scales: 5 warning signs
Mornings lost when the internet drops, broken meetings, new starters waiting a week for access: the signs of an infrastructure that grew through patches.
More than 10 workstations · support contract · remote, on-site and phone support
02 · A network infrastructure that no…
The server lives at its limit. Do I buy more disk?
That does not solve it. Disk at 90%, memory exhausted mid-morning, ERP slow at peak time: adding capacity postpones the problem by about six months. The cause is a server designed for the size the company was five years ago.
Because it was designed for a different company. In the morning it works; from 10am it degrades; in large meetings video calls drop. The internet is not to blame. The wireless network was built for 15 people and today has to serve 50 users with 120 devices.
It is the most visible symptom. Unlabelled cables, cascaded switches of several brands, the ISP router acting as a firewall and an unwritten rule: "don't touch". Every fault takes hours just to locate, because nobody knows the connection map.
01Patched-together infrastructure costs more in downtime than redesigning it would.
02The five signs: server, Wi-Fi, rack, onboarding new staff and a single internet line.
03Start with redundant internet and the rack: fast, cheap and with the biggest impact.
04Velv by OnRising grew from 90 to 200 workstations without changing architecture.
In this article
A network infrastructure that has stopped scaling charges you every day: mornings lost when the internet goes down, meetings cut off by the Wi-Fi, new employees waiting a week for access. The business grows in leaps, a dozen hires in three months, new premises, new software, and IT keeps up with patches: one switch hanging off another, one more disk, a retail access point. It works for a few years. Then the symptoms appear, and ignoring them always costs more than dealing with them. These are the five we find most often.
The server lives at its limit. Do I buy more disk?
That does not solve it. Disk at 90%, memory exhausted mid-morning, ERP slow at peak time: adding capacity postpones the problem by about six months. The cause is a server designed for the size the company was five years ago.
The right decision is strategic, replace, virtualise or move to the cloud, and should look at the company three years from now. We compare the three options in replace the server, virtualise or go to the cloud. With Windows Server 2016, the deadline is already set: January 2027.
Why does the Wi-Fi fail when the room fills up?
Because it was designed for a different company. In the morning it works; from 10am it degrades; in large meetings video calls drop. The internet is not to blame. The wireless network was built for 15 people and today has to serve 50 users with 120 devices.
It is the most visible symptom. Unlabelled cables, cascaded switches of several brands, the ISP router acting as a firewall and an unwritten rule: "don't touch". Every fault takes hours just to locate, because nobody knows the connection map.
It is also the cheapest to fix relative to the hours it wastes:
Organise The rack and label every connection.
Document The topology.
Replace the cascade With stacked managed switches.
What does an employee who can only start working after a week cost?
A great deal, and it multiplies with size. Email on day two, laptop on day three, ERP access the following week: with no standard image or process, every workstation is configured by hand and every access depends on someone remembering. With 15 people it is tolerable. With 60, lost productivity adds up to weeks every year.
It is solved with three pieces: a standardised workstation image, role-based access profiles that generate the accounts and a one-page onboarding procedure, the same model as the 7 cybersecurity policies we recommend.
What happens when the only internet connection goes down?
The company stops completely. Email on Microsoft 365, ERP in the cloud, IP phones: everything depends on one line, and only the operator can restore it. That was tolerable with systems in-house; it stopped being so with everything in the cloud.
A backup line, mobile network or a second operator, with automatic failover on the firewall is cheap compared with a morning without work. We did it at FreshDesign, with load balancing and automatic failover, read the FreshDesign case study. The decision is analysed in why internet speed is not enough.
What is the sign behind all the others?
Lack of visibility. If counting workstations, licences, servers and access points takes more than 24 hours, the infrastructure is not being managed, it is being patched. Before spending money on any of these symptoms, take the inventory. The IT estate audit checklist can be done in an afternoon.
How do I design an infrastructure for growth?
Network and Wi-Fi for double the size The price difference is small; the difference in service life is large.
Servers after the strategy Define what moves to the cloud and what stays in the office; buying comes afterwards.
Documentation Of everything that is installed.
Preventive monitoring Alerts before the limit is reached, not after.
For IT Unlimited clients, 24×7 NOC monitoring, included in Premium 365, continuously tracks disk and memory usage and the behaviour of the network and Wi-Fi; renewal is planned once a year with the client. That is how Velv by OnRising grew from 90 to 200 workstations while keeping the same architecture.
Questions decision-makers ask
For how many years should infrastructure be sized?
Cabling and network for 10 years or more, because the building work is the most expensive part and the extra cost is low. Servers and Wi-Fi for four to five years; workstations are renewed over the same interval.
Does moving everything to the cloud solve the problem?
Not on its own. It removes the server capacity issue, but makes the company more dependent on the internal network and the external link. Without redundancy and solid Wi-Fi, it simply moves the problem elsewhere.
What investment does modernising the network of a 50-person office require?
It depends on the existing cabling and what can be reused, so only a site visit can pin down the price. It is almost always less than what the company loses in a year to faults and downtime.
If I have all five signs, where do I start?
Start with the rack and the second internet line. They are quick, cheap jobs with the biggest impact on downtime. Wi-Fi and the server come next, with a proper design.
Want to know what is at its limit in your company?
We analyse the infrastructure, flag in a report what can no longer cope and design a phased plan, with the cost of each stage in plain sight, with no commitment. Discover DataRoad's managed IT services.
More than 10 workstations · support contract · remote, on-site and telephone support. +351 211 459 950 * · sales@dataroad.pt