top of page
Search

Network Capacity Planning for Growing Offices and Stores

Writer: tekmatik303
tekmatik303
Nov 12
11 min read

Abstract illustration of planned network capacity

Network capacity planning means measuring what your network carries today, forecasting what it will carry after you add people, devices, cameras, registers or locations, and upgrading before users feel the slowdown. For a business with one to a few sites, the work comes down to four things: a two-week usage baseline, a growth forecast, a failure check, and an upgrade order that starts with the real bottleneck. We install and maintain business networks in Denver, so this guide follows the order we would use on a site walk, and it points out where cabling, switches and Wi-Fi usually decide the outcome.



TL;DR

 

  • Capacity is more than an internet speed number. Demand, topology, traffic type and what happens when a link fails all have to be looked at together, as Cisco’s planning method does.

  • Measure first. Collect about two weeks of utilization data so regular peaks, quiet periods and recurring spikes all show up in the baseline.

  • Forecast with lead time in mind: if an upgrade takes three months to deliver, you need to plan at least four months ahead.

  • Size for the bad day, not the average day. Traffic that reroutes during a failure needs somewhere to go.

  • Expand where the data shows congestion, not everywhere at once. The first fix is often a switch, a cable run or an access point, not a bigger internet plan.

 

Table of Contents

 

 

What network capacity planning covers in a business

 

Capacity is the maximum amount of data your network can move between two points. Planning is the habit of checking that capacity against demand before the two collide. Cisco’s guidance on availability puts the purpose plainly: capacity planning determines the resources needed to prevent performance or availability impacts on business-critical applications. In an office or store, those applications are the ones you notice when they fail: the point-of-sale terminal that hangs at checkout, the video call that freezes, the cloud accounting tool that crawls, the camera feed that drops frames.

 

It helps to think of the network as several layers, each with its own ceiling:

 

  • The internet connection. The bandwidth you buy from your provider, in both directions.

  • The core and switches. The equipment that connects every wired device, and the uplinks between switches.

  • The cabling. The physical runs that set the top speed a device can negotiate. Our business network cabling guide covers the choices that keep a building ready for growth.

  • Wi-Fi. Radio capacity is shared among everyone in range, so ten more laptops in one conference room is a different problem than ten more on a different floor.

  • Applications. Voice, video, point-of-sale traffic, cloud backup, security cameras and software updates all compete for the same pipe.

 

A slowdown can come from any layer. That is why a bigger internet plan sometimes fixes nothing: the bottleneck was a switch uplink or an overloaded access point.

 

Step 1: Take stock of what is on the network

 

Before you measure traffic, list what generates it. A short inventory is enough for a small business:

 

  1. Every wired and wireless device by type: workstations, laptops, phones, printers, point-of-sale terminals, cameras, access control panels, displays, tablets, guest devices.

  2. Every switch, router, firewall and access point, with its port speeds and uplink speeds.

  3. Every site, and how each one reaches the internet and any other site.

  4. The business-critical applications and who uses them at what hours.

 

ManageEngine’s monitoring checklist starts the same way: discover the network elements, then establish baselines. An inventory also exposes the quiet problems, such as a 24-port switch with 23 ports in use, an uplink running at a tenth of what the hardware supports, or cameras added to the office network with no plan for their traffic.

 

Step 2: Measure your baseline

 

A baseline is a record of normal. Without it, you cannot tell whether a complaint is a capacity problem, a fault, or a one-off.

 

What to collect. Track capacity, inbound and outbound utilization, throughput and traffic trends on each important link. Also track latency (delay), jitter (variation in delay), packet loss, errors and discards, plus which applications and devices send the most traffic. ManageEngine’s router monitoring guide recommends watching inbound and outbound utilization separately, because WAN traffic is often asymmetric. Uploading backups and video calls load the upstream side far more than browsing does.

 

How long to collect. NinjaOne suggests two weeks of data can cover regular activity, highs, lows, recurring spikes and quiet periods, and advises leaving out unusually high or low days from the baseline. Avoid a holiday week or a week when a system was down. A retailer should include a weekend and a promotion day. An office should include month-end, when accounting and reporting jobs tend to run.

 

How to collect it. TechTarget recommends using SNMP (a standard way for network gear to report its own statistics) or a modern analytics platform to review historical bandwidth, device CPU, memory, usage and performance. Most business-grade routers, switches and firewalls expose these counters. The hard part is rarely getting the data. It is knowing which numbers matter and reading them together.


Step 2: Measure your baseline

Read utilization with symptoms, not alone

 

A link at 85% utilization is not automatically a problem. ManageEngine points out that high utilization without degradation is different from high utilization with delivery problems. Check packet loss, latency, jitter, errors and discards at the same time. If a link is busy but clean, you have headroom concerns to plan for. If it is busy and dropping packets, users are already paying for it. If errors and discards are high on a link with low utilization, the cause is probably a cable, a port or a duplex mismatch. That is a repair, not an upgrade. Our guides on network troubleshooting for business downtime and what causes intermittent network outages help separate the two.


Network symptoms distinguish capacity needs from repairs

Step 3: Sort traffic by what it is worth to the business

 

Two links with the same utilization can deserve very different decisions. Intel’s WAN capacity management paper reports that classifying traffic by type helped separate lower-priority backup spikes from business-critical interactive, design and videoconferencing demand, which improved its bandwidth-investment decisions.

 

The same logic applies at small scale. Sort what you find into three groups:

 

Traffic type

Examples

How it behaves

Planning question

Business-critical, real-time

Voice calls, video meetings, card payments, POS lookups

Small to moderate volume, intolerant of delay and loss

Is it protected from everything else?

Business-critical, bulk

Cloud file sync, remote desktop, ERP and inventory updates

Heavy and steady, tolerates some delay

Does it fit inside business hours?

Lower priority

Offsite backups, software updates, guest Wi-Fi, streaming

Heavy, bursty, can wait

Can it be scheduled or limited?

Often the cheapest capacity you will ever gain is moving a nightly backup from 2 p.m. to 2 a.m., or limiting guest Wi-Fi so it cannot starve the register. Phone traffic is a good place to start because its needs are predictable. For a rough sense of what voice demands per call, see these VoIP bandwidth requirements when you add up your real-time traffic.

 

Cameras deserve their own line in the inventory. A security camera system sends continuous video across the network, and each camera you add changes the load on its switch and uplink. Planning cameras, access control and the data network together keeps the video from crowding out the register. Coordinating those projects is part of what we do.

 

Step 4: Forecast demand with lead time built in

 

A forecast starts from your baseline and adds what the business plans to change. A backbone planning methodology paper describes the core method: analyze historical traffic, project growth with trendlines and growth rates, and account for acquisitions, new applications, product offerings and network changes. It also notes that input from business groups can help determine additional headroom.

 

For an office or store group, translate that into questions you can answer in a meeting:

 

  • How many people or registers will you add in the next 12 to 24 months?

  • Are you opening, moving or closing a location?

  • Are you adopting video meetings, cloud phone service, cloud backup, new POS software or more cameras?

  • Will devices per person rise (laptop, phone, tablet, badge reader, wearable)?

  • Is there a seasonal peak, such as a holiday sales period?

 

A worked example

 

Suppose your internet link peaked at 220 Mbps downstream during the baseline period, and you expect staff and cloud use to grow about 10% over the planning period. Cisco’s whitepaper describes growth as a multiplier: a 10% expected increase multiplies current demand by at least 1.1, so the forecast peak is about 242 Mbps. Cisco’s WAE tool supports compounded growth rates per period, so a 10% yearly increase over two years is 220 x 1.1 x 1.1, about 266 Mbps. That is demand, not what you should buy. The overprovisioning step comes next.

 

Plan for how long an upgrade takes

 

The forecast horizon has to cover delivery time. Cisco’s example: if an upgrade takes three months and you plan monthly, forecast at least four months ahead. For a business, delivery time includes the internet provider’s install window, a cabling run that needs a permit or landlord approval, and hardware shipping. If you wait until users complain, you wait out the lead time with a slow network.

 

More sites add more variables. If you are adding locations, our guide on multi-site rollout planning covers how to sequence the work.

 

Step 5: Add headroom and size for failures

 

Cisco’s method does not stop at forecast demand. It applies an overprovisioning factor, the margin that keeps service quality acceptable at peak, and then simulates the forecast load on the topology, including failure cases. The point, as the backbone paper stresses, is that resilient capacity has to handle rerouted traffic during a failure, not just normal steady-state load.

 

For a business, this turns into plain questions:

 

  • If one switch uplink fails, does the other carry everything?

  • If the primary internet circuit goes down, can a backup connection run the register, phones and cloud applications at an acceptable level?

  • If one access point dies, do nearby ones absorb the clients?

 

The backbone paper tests link failures, and sometimes node failures, and then sizes each surviving link for the largest requirement among the scenarios. You can do a lightweight version: list your single points of failure and write down what each one would push onto the remaining paths. Our overview of business internet backup solutions explains the options for a second path, and the retail network outage case study shows what a single failure looks like across several stores.

 

Step 6: Expand capacity where the data points

 

Once the baseline, forecast and failure check are done, you know which layer is short. Match the fix to the layer:

 

Where the limit is

Typical signs

Typical expansion

Internet connection

High utilization with latency and loss at peak, in or out

Higher-tier plan, second circuit, scheduling bulk traffic

Switch uplinks or core

Busy trunk links, drops between floors or rooms

Faster uplinks, link aggregation, newer switches

Cabling

Errors on a link, devices stuck at low negotiated speeds

Recabling the affected runs, new drops for added devices

Wi-Fi

Complaints in dense areas, good wired performance

More access points, better placement, moving fixed devices to wired

Specific applications

One app starts the problem

Prioritization, scheduling, relocating the workload

Two principles from the research help here. First, expand progressively. Nokia describes a modular optical approach in which new equipment goes first onto the most congested links, so capacity grows without upgrading every link at once. In an office, that means upgrading the one congested uplink, not the whole building. Second, check physical feasibility before you order. Microsoft’s documentation on upgrading circuit bandwidth notes that an upgrade may need a new circuit if the physical port lacks capacity, and recommends validating the new circuit on a test network before switching production over. The lesson carries to any site: the plan you can buy is limited by the hardware and cabling it lands on.

 

Wired versus wireless is often the deciding call when you add devices. Fixed devices such as registers, desktops, printers and cameras belong on cable, which frees Wi-Fi for mobile devices. Our comparison of wired vs wireless networking for business goes through the tradeoffs.

 

Finally, tie the technical plan to budget and procurement. Fastly describes monthly aggregation of planning data feeding forecasts, buildouts, capital and operating budgets, and coordination with hardware supply. You do not need a monthly cycle with that rigor. You do need a date on the calendar when someone looks at the numbers and decides.

 

Step 7: Keep the plan current

 

A capacity plan is a living document. Update the network model with planned changes and upgrades, apply growth projections to the latest traffic data, and rerun the failure checks. In practice:

 

  • Review utilization quarterly, and sooner when you add a department, a store, a camera system or a new cloud application.

  • Keep alerts on utilization, errors and discards so a trend shows up before users report it.

  • Re-baseline after any major change, since the old normal no longer applies.

  • Keep switches, firmware and cabling documented, so the next assessment starts from facts.

 

Routine upkeep supports all of this. Our piece on network maintenance that keeps business moving lays out what regular care looks like.

 

How Tekmatik helps with capacity planning

 

The measurements and forecasts above are manageable. The physical work that follows is where most plans stall: pulling new cable runs, replacing switches, adding access points and reworking a rack, all in a space people are still using. Tekmatik provides structured cabling, commercial network setups, office Wi-Fi and workstation deployment, CCTV, access control, POS system installation, and ongoing troubleshooting, maintenance and repair. Because we work on both IT and physical security systems, we can coordinate a network expansion with the cameras or access control that depend on it. Our installation approach is built around precision, minimal disruption and a clean site when we leave.

 

A good first step is an initial phone consultation to talk through what you are seeing. If you want someone on the premises to look at the cabling, switches and Wi-Fi, we offer an on-site assessment for a one-off fee of 99 USD. Details are on the consultation page, and larger or multi-site needs can be discussed under our Enterprise Plans.


Tekmatik

Our perspective

 

Most capacity complaints we hear start as an internet speed question. The real answer is usually spread across the building: a cable run that was never meant for the devices on it, a switch carrying more than it was sized for, an access point serving a room it was not placed for.   That is why we treat capacity as a premises problem as much as a bandwidth problem. Cabling, switching, Wi-Fi, cameras, access control and POS all share the same infrastructure, and a plan that looks at only one of them tends to move the bottleneck rather than remove it.   We believe the best plan is the one you can actually carry out without stopping work. Our installations are done with care for your operations and your space, and our support continues after the install, so the plan gets revisited as your business grows.   Tekmatik

 

FAQ

 

How much spare capacity should a business keep?

 

There is no single figure that suits every site. Cisco’s method applies an overprovisioning factor chosen for the service level you need, on top of forecast demand. A practical approach is to set headroom from your baseline peaks, your growth forecast and the delivery time of an upgrade, then confirm the result against a failure scenario.

 

How often should we review our capacity plan?

 

Review it on a regular schedule, such as quarterly, and after any major change like a new location, a headcount jump, or new cameras or POS systems. Re-baselining after changes keeps the numbers honest. Alerts on utilization, errors and discards help you catch trends between reviews.

 

Is a faster internet plan the first thing to try?

 

Only if the data shows the internet link is the constraint, meaning high utilization together with latency or packet loss at peak. Busy links with clean performance are a planning item, and problems with low utilization often come from cabling, ports or Wi-Fi. Measure first, then match the fix to the layer.

 

What do we need to start?

 

Start with a device and equipment inventory and about two weeks of utilization data from your router, switches and firewall. Add your plans for headcount, locations, cameras and applications. If you would rather have someone review the site with you, a Tekmatik phone consultation or on-site assessment is a straightforward way to begin.

 

Sources

 

Recommended

 

 
 
 

Comments


bottom of page