Recommended Posts

Hi everyone,

I'm planning a small network upgrade and was wondering how others prepare their networks for future needs.

Do you usually invest in higher-speed switches and better cabling from the start, or do you upgrade only when necessary?

I'd be interested in hearing what has worked well for you and any lessons you've learned over time.

Thanks!

Depends on what you need. Might be a bit clearer on what you plan to do with it.

Sort of a waste if you get the newest and greatest, but don't know how to use it.

I hate the term, "future-proof." We saw it back in the 90's / 2000's, if not before. You cannot future-proof anything, since there is no definition of how far into the future you plan on prepping for.

Best idea is to tell us what you currently have and what its use is at the moment, and we can then offer ideas about some areas that might need an upgrade and other areas that can be left alone.

On 29/06/2026 at 06:00, rosiecharles said:

Hi everyone,

I'm planning a small network upgrade and was wondering how others prepare their networks for future needs.

Do you usually invest in higher-speed switches and better cabling from the start, or do you upgrade only when necessary?

I'd be interested in hearing what has worked well for you and any lessons you've learned over time.

Thanks!

Personally, I’ve found that it’s usually worth investing in the infrastructure you don’t want to replace later, especially cabling. Running Cat6A (or better, depending on your needs) during an upgrade is relatively inexpensive compared to having to re-cable a few years down the road.

For switches I try to balance current specs with realistic growth. If my budget allows it Ill choose switches with higher uplink speeds which leaves room for expanding later on, but I don’t necessarily overspend on access ports if the endpoints won’t benefit from them anytime soon.

One lesson I’ve learned is that planning for scalability pays off. It’s much easier to add devices, VLANs, or higher-bandwidth workloads when your network infrastructure already supports it than to replace hardware later. 

What is your budget like?

When I think about a network, there are really two aspects, the hardware and the wiring. So here is what I would do for both.

Wiring:

  • Use Cat6A for the patch panel, outlets, and all structured cables (cables installed in walls).
  • Run plenty of Wireless Access Point (WAP) cables, as a general rule, assume a signal can only pass through 2-3 walls and can't pass through a floor (that is conservative, but trust me on this if you want strong WiFi) 
  • Cat6 patch cables are fine for now if you don't plan to run 10gig, those are easy to replace later if needed.
  • Run OS2 single-mode fiber to anywhere you think you may have a server or sub-switch. (yes, single-mode for everything on a small network, don't mess with multimode unless you are at a scale where that minor cost and power savings will matter).
  • If you really want to future proof, also run fiber to any high density WAP locations, it is likely that WiFi 8 and beyond WAPs will push the limits of 10g.
  • Run 6-12 pairs of single-mode fiber between your MDF and the building's MDF, even if you only need 1 or 2 pairs now, those extra pairs will pay off down the road.

Hardware: (its easy to say "get all the features incase you need them", so instead of futureproofing, I am going to take approach of suggesting areas worth investing in, and areas you can save money).

  • Don't overspend thinking you need every feature on every port. You don't need 10g on every port, you don't need PoE on every port.
  • Don't overspend on redundancy either, unless you are ready to buy two of everything, don't waste money buying two of some things and not others. Dual power supplies are worthwhile, but probably not HA or multi-path redundancy. 
  • Get 1 "distribution layer" switch that your router/firewall will connect to as well as all your access layer switches below. This should be a fully managed 10g+ switch with a combination of copper and SPF ports, a few 25g uplink ports are nice for this switch.
  • Given that you said it is a small network, I suggest also using that distribution layer switch for servers and WAPs, meaning it will need PoE.
  • Speaking of wireless, get good professional tri-band WAPs, and either turn on the band stirring options, or limit 2.4 to an IoT only SSID. This will provide a solid WiFi capable nearly everything but the highest of bandwidth clients...you could even consider skipping wiring workstations depending on usage.
  • Access layer switch for workstations and printers can be cheaper switches, 2.5g is a good sweet spot between price and future proofing, but even 1g is fine for most individual clients. These switches are easy to replace later, so don't pay for 10g ports if you don't need them now, they will likely be a lot cheaper in 5 years.
  • You can consider saving a little on access layer switches by only getting 1 PoE switch for whatever needs it (remember your WAPs are connecting to the distribution switch, not here), and non-PoE for your workstations, because desk phones are falling out of favor.
  • You can also save money here by not buying managed switches if you don't need them--but really do some soul searching there, if you go this route, then anything that isn't on your workstation VLAN would either need to be connected to the distribution switch, or its own access layer switch.
  • Also, don't feel like you need a fancy fabric stacking switches for your access layer, that is the point of the higher-end distribution layer, to remove the need for things like that at this level.

Home Hardware:

  • I'm realizing the above assumed an office setting, if this if for your house and home lab then the above still applies, but you'll probably want everything managed and PoE, just because, but you probably also don't need multiple access layer switches.
  • If your total port count is below 24, just skip separating distribution layer and access layer and just get one nice switch with the features you want.
  • If you are at the point of considering a 48-port switch, I would instead get a nice high-end distribution switch for things that need it, and cheaper access layer switches with specs based on the needs of connected devices.
  • For home use, don't worry about home running every device to the main switch, there is nothing wrong with running sub-switches for your media areas and office, those essentially become your access layer, just look for sub-switches with a 10g uplink so sharing bandwidth isn't an issue. Just make sure you always connect them to your distribution/main switch, don't daisy chain, the path should never have more steps than Client>Access>Distribution>Firewall>Internet or Client>Access>Distribution>Server if it is local.
  • Love 2
On 29/06/2026 at 18:49, Mindovermaster said:

Depends on what you need. Might be a bit clearer on what you plan to do with it.

Sort of a waste if you get the newest and greatest, but don't know how to use it.

Thanks for the feedback. That's a fair point.

This is mainly for a small home office setup with a few wired devices, Wi-Fi for laptops and mobile devices, and room for a NAS in the future. I'm trying to avoid replacing the core networking equipment too soon while keeping the budget reasonable.

I appreciate your advice.

On 29/06/2026 at 20:59, Nick H. said:

I hate the term, "future-proof." We saw it back in the 90's / 2000's, if not before. You cannot future-proof anything, since there is no definition of how far into the future you plan on prepping for.

Best idea is to tell us what you currently have and what its use is at the moment, and we can then offer ideas about some areas that might need an upgrade and other areas that can be left alone.

I understand what you mean, and that's a good point.

I probably should have said "planning for future upgrades" rather than trying to future-proof everything. My goal is simply to avoid making decisions now that I'll regret in a couple of years.

Thanks for pointing that out.

On 29/06/2026 at 23:56, sphbecker said:

When I think about a network, there are really two aspects, the hardware and the wiring. So here is what I would do for both.

Wiring:

  • Use Cat6A for the patch panel, outlets, and all structured cables (cables installed in walls).
  • Run plenty of Wireless Access Point (WAP) cables, as a general rule, assume a signal can only pass through 2-3 walls and can't pass through a floor (that is conservative, but trust me on this if you want strong WiFi) 
  • Cat6 patch cables are fine for now if you don't plan to run 10gig, those are easy to replace later if needed.
  • Run OS2 single-mode fiber to anywhere you think you may have a server or sub-switch. (yes, single-mode for everything on a small network, don't mess with multimode unless you are at a scale where that minor cost and power savings will matter).
  • If you really want to future proof, also run fiber to any high density WAP locations, it is likely that WiFi 8 and beyond WAPs will push the limits of 10g.
  • Run 6-12 pairs of single-mode fiber between your MDF and the building's MDF, even if you only need 1 or 2 pairs now, those extra pairs will pay off down the road.

Hardware: (its easy to say "get all the features incase you need them", so instead of futureproofing, I am going to take approach of suggesting areas worth investing in, and areas you can save money).

  • Don't overspend thinking you need every feature on every port. You don't need 10g on every port, you don't need PoE on every port.
  • Don't overspend on redundancy either, unless you are ready to buy two of everything, don't waste money buying two of some things and not others. Dual power supplies are worthwhile, but probably not HA or multi-path redundancy. 
  • Get 1 "distribution layer" switch that your router/firewall will connect to as well as all your access layer switches below. This should be a fully managed 10g+ switch with a combination of copper and SPF ports, a few 25g uplink ports are nice for this switch.
  • Given that you said it is a small network, I suggest also using that distribution layer switch for servers and WAPs, meaning it will need PoE.
  • Speaking of wireless, get good professional tri-band WAPs, and either turn on the band stirring options, or limit 2.4 to an IoT only SSID. This will provide a solid WiFi capable nearly everything but the highest of bandwidth clients...you could even consider skipping wiring workstations depending on usage.
  • Access layer switch for workstations and printers can be cheaper switches, 2.5g is a good sweet spot between price and future proofing, but even 1g is fine for most individual clients. These switches are easy to replace later, so don't pay for 10g ports if you don't need them now, they will likely be a lot cheaper in 5 years.
  • You can consider saving a little on access layer switches by only getting 1 PoE switch for whatever needs it (remember your WAPs are connecting to the distribution switch, not here), and non-PoE for your workstations, because desk phones are falling out of favor.
  • You can also save money here by not buying managed switches if you don't need them--but really do some soul searching there, if you go this route, then anything that isn't on your workstation VLAN would either need to be connected to the distribution switch, or its own access layer switch.
  • Also, don't feel like you need a fancy fabric stacking switches for your access layer, that is the point of the higher-end distribution layer, to remove the need for things like that at this level.

Home Hardware:

  • I'm realizing the above assumed an office setting, if this if for your house and home lab then the above still applies, but you'll probably want everything managed and PoE, just because, but you probably also don't need multiple access layer switches.
  • If your total port count is below 24, just skip separating distribution layer and access layer and just get one nice switch with the features you want.
  • If you are at the point of considering a 48-port switch, I would instead get a nice high-end distribution switch for things that need it, and cheaper access layer switches with specs based on the needs of connected devices.
  • For home use, don't worry about home running every device to the main switch, there is nothing wrong with running sub-switches for your media areas and office, those essentially become your access layer, just look for sub-switches with a 10g uplink so sharing bandwidth isn't an issue. Just make sure you always connect them to your distribution/main switch, don't daisy chain, the path should never have more steps than Client>Access>Distribution>Firewall>Internet or Client>Access>Distribution>Server if it is local.

Thanks for taking the time to write such a detailed explanation.
I hadn't considered running fiber for future expansion, so that's something I'll definitely read more about.

How many users? What type of traffic? Nature of the building?

@sphbecker covered it well, get Cat6 everywhere you will want to put either desks, or equipment such as shared printers - cabling is the same whatever you do and it's the cabling that gives you the options in the future.


But in terms of equipment, you seem to be very much the smaller end of the spectrum, so you're unlikely to require true core networking.
If you're sub 100 seat I'd be getting a decent firewall and doing 'core' networking on that, if you have requirements for VLANs etc any decent NGFW would be where I'd do the L3 work, leaving you only requiring L2 switches downstream.
10Gb - sure it's nice, but depends on what you do - if you're an 100 user accountants office 1Gb is more than sufficient to get excel in front of your users and will be until the end of time, if you're 10 user video editing studio that's a different proposition and 10Gb or faster might well be worth the money.

Ubiquiti is not my thing personally, but it's hard to argue with the value and featureset in the SMB space.

  • 2 weeks later...

I usually future-proof the parts that are expensive to replace later. If I'm running new cables, I always choose Cat6 or Cat6a instead of the cheapest option. For switches, I buy one with a few extra ports so I have room to expand. I don't always chase the latest networking standard unless I actually need it. I've found that a reliable router and good Wi-Fi coverage make a bigger difference than raw speed for most home networks. Planning ahead has saved me from doing a complete network rebuild later.

You can easily go overboard here. From what you are saying, it is a home office, a future NAS, a few client devices and one maybe two wireless access points. Many of the above tips are great and completely spot on, but are aimed at larger setups. If you are interested in networking and want to play or want to pursue a networking interest for your career then exposure to the tips above is a good idea.

If however the network is something you want to largely get right and forget about for 5-10 years then for your home you can dial it back.

 

Perhaps some questions to help focus in on requirements:

How many wired network connections do you need now? Count up all the devices. Assume for a SOHO NAS you'll be using two network ports. This - with a sensible growth factor - defines the minimum switch size.

Do you have any idea how many wireless access points you'll need to give the coverage of your house/flat that you want? Might you move within the life of this equipment and need to change "everything"? Or might you want coverage over multiple floors/out into a garden? Do you currently have any signal problems or dead zones?

Is there any interest in adding CCTV now or in the future?

Are you a heavy IoT/wi-fi smart home device user or do you avoid these things?

Are you in a flat, or a bungalow, two-story, three-story house? Will you need to get many wired connections between floors or is everything concentrated with only a couple of wireless access points in the upstairs/basement spaces?

 

In terms of cost savings

If you are not certain on wireless access point densities, but have a scaled floor plan to hand, try plugging it into the Ubiquiti designer -https://design.ui.com/ (Create project in the bottom right, then find the floor plan option in the menu and upload the image(s). Don't forget to set the scale before drawing in some walls and dragging a couple of  AP's in to see what you might expect to receive).

You will want to get 2.5GbE ports, but not necessarily all 2.5GbE. Modern wireless access points increasingly use 2.5GbE even if your equipment is only on 1GbE. 5GbE might be worth considering, but will change the price point. You can pick up very inexpensive 2.5GbE / 5GbE USB adapters and PCIe adapters for your larger devices if you don't have anything over 1GbE (you can even use them with some SmartTV's which tend to only be 100Mbps too). The same for replacing old Wi-fi chips in laptops and micro PCs (they tend to be replaceable) and this will give you more incentive to push for better equipment out the gate.

As has been said by others, if you are pulling cables through walls, floors or ducting. Spend now and don't regret it later. If you are making outdoor runs, invest in the correct shielded cabling now and ensure that it is earthed. Cat6 will run 10GbE provided that the run distance isn't excessive. If you need the full 100m run though, go for 6a. Keep in mind that 6a is stiffer, heavier, harder to terminate and more expensive, so you could compromise with 6a between floors and then Cat6 into rooms.

Get a decent router. Get away from the junk your ISP will provide you with. It will give you options and will last a LONG time. You will be able to move ISP without second thought if you have a decent router as a base. If you are techy and want all the toys and options, something like a NetGate appliance / a pfSense Community self build (there are plenty of reasonably cheap small form factor chassis you can pickup for this on Amazon etc). If you just want something to work and be easy to configure and find help for though, Ubiquiti.

Whether you need Power Over Ethernet (PoE) depends upon your appetite for PoE devices. If you need more than a couple of PoE devices, then a switch is worth considering - CCTV, access points, door security. If you have zero interest in any of that and will only wind up with one or two access points in the long run, and IF you are able to put them near 120/240v power, then a couple of PoE injectors and plugging them into the wall will save you money. If you change your mind later and start accumulating PoE devices though, you will regret not getting PoE now.

Space: large switches with high port densities, PoE, redundancy, SFP ports (for fibre optics or 10/25GbE) quickly become big devices. If you don't have a comms room in mind and don't want them on display in your home, then keep in mind that you might not want to start exploring prosumer and SMB networking equipment for your home. 

 

If you can offer us some of the answers to these questions and your thinking on needs, perhaps we can help you drill into the 'what' 🙂

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Posts

    • They've already bought Clipchamp, replaced Movie Maker with it, and made it worse in the end 🤣
    • Microsoft doesn't label their download like 2026.09.07. Their page only says updated on July 9, 2026. I've never seen anybody use YYYY-DD-MM before. When I saw this article I read it as a September 7th update. Shouldn't it be YYYY-MM-DD, like 2026.07.09?
    • This innovation could solve Global Warming and change refrigerators forever by Sayan Sen Image by Tuan Vy via Pexels Researchers at the U.S. Department of Energy’s Lawrence Berkeley National Laboratory (Berkeley Lab) developed a new way of heating and cooling that used electrically charged atoms or molecules, known as ions, to control changes between solid and liquid states. The method, called "ionocaloric cooling," was described in a study published in the journal Science. The idea was based on a familiar example. When salt is spread on roads before a winter storm, it changes the temperature at which ice forms. The new method worked in a similar way by using ions to change the melting point of a material. This created a controlled phase change, which is the process of a material changing between solid and liquid. When the material melted, it absorbed heat from its surroundings. When it became solid again, it released that heat. This allowed heat to be moved from one place to another without using conventional refrigerant gases. The study addressed a long-standing challenge in refrigeration: developing cooling systems that were energy efficient while using refrigerants that had little or no impact on global warming. Most refrigerators and air conditioners use vapor-compression refrigeration, where a refrigerant gas is compressed and expanded in a continuous cycle to absorb heat from one place and release it somewhere else. Many of these systems use hydrofluorocarbons (HFCs), synthetic refrigerant gases with a high global warming potential (GWP), a measure of how much heat a greenhouse gas can trap in the atmosphere compared with carbon dioxide. The ionocaloric system instead used only solid and liquid materials, removing the need for refrigerant gases. “The landscape of refrigerants is an unsolved problem: No one has successfully developed an alternative solution that makes stuff cold, works efficiently, is safe, and doesn’t hurt the environment,” said Drew Lilley, a graduate research assistant at Berkeley Lab and PhD candidate at the University of California, Berkeley, who led the study. “We think the ionocaloric cycle has the potential to meet all those goals if realized appropriately.” Heating and cooling accounted for more than half of the energy used in homes, making improvements in this area important for reducing energy use and greenhouse gas emissions. Replacing current refrigerants was also part of international climate efforts, including the Kigali Amendment, which was accepted by 145 parties, including the United States in October 2022. Under the agreement, countries committed to reducing the production and consumption of HFCs by at least 80% over the next 25 years. Ionocaloric cooling was one of several caloric cooling technologies under development. These systems worked by making materials absorb or release heat when exposed to an external force. Some used magnetic fields, while others used electric fields. However, these approaches often required relatively strong applied fields while producing smaller temperature changes and lower efficiency. According to the researchers, ionocaloric cooling could produce larger temperature and entropy changes using much lower applied field strengths. In thermodynamics, entropy describes how energy is distributed within a system and plays an important role in how heat moves. Because the working material became a liquid during the process, it could also be pumped through a system, making it easier to transfer heat than in many solid-state cooling technologies. Lilley and Ravi Prasher, a research affiliate in Berkeley Lab’s Energy Technologies Area and adjunct professor of mechanical engineering at the University of California, Berkeley, developed the theory behind the ionocaloric cycle. Their calculations suggested that the method could match or even exceed the efficiency of conventional refrigerant-based cooling systems. To test the concept, the researchers built an experimental system based on an ionocaloric Stirling refrigeration cycle, a version of the Stirling thermodynamic cycle adapted to move heat using ion-driven phase changes instead of conventional refrigerants. The system used a sodium-iodine salt together with ethylene carbonate, an organic solvent commonly used in lithium-ion batteries. Applying a small electrical current moved ions through the system, changing the material's melting point. As the material melted, it absorbed heat. When the ions were removed, the material solidified and released the stored heat. The researchers reported that the experimental system achieved a coefficient of performance (COP) equal to about 30% of the theoretical Carnot limit, which represents the highest possible efficiency any cooling system can achieve under ideal conditions. It also produced an adiabatic temperature change, meaning the material's temperature changed without gaining or losing heat to its surroundings, of up to 25 degrees Celsius using an applied voltage of about 0.22 volts. According to the study, this temperature change was larger than those demonstrated by other caloric cooling technologies while operating at a comparatively low voltage. “There’s potential to have refrigerants that are not just GWP [global warming potential]-zero, but GWP-negative,” Lilley said. “Using a material like ethylene carbonate could actually be carbon-negative, because you produce it by using carbon dioxide as an input. This could give us a place to use CO2 from carbon capture.” Prasher said the team was trying to balance environmental impact, energy efficiency and equipment costs. “There are three things we’re trying to balance: the GWP of the refrigerant, energy efficiency, and the cost of the equipment itself,” Prasher said. “From the first try, our data looks very promising on all three of these aspects.” While most caloric technologies were discussed for cooling, the researchers said they could also be used for applications such as water heating and industrial heating. The team was continuing to develop prototypes to test different materials, improve the system's efficiency and temperature range, and determine whether the technology could be scaled up for practical use. “We have this brand-new thermodynamic cycle and framework that brings together elements from different fields, and we’ve shown that it can work,” Prasher said. “Now, it’s time for experimentation to test different combinations of materials and techniques to meet the engineering challenges.” Source: Berkeley Lab, Science This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing
    • Microsoft has several different customer groups. Services (like Azure), Office, dev tools, Windows server and Windows client which itself breaks into three groups: enterprise, education and consumer. The consumer product is one of the least profitable sections and is mainly used to get consumers to buy the other groups like Office and Services, so MSFT tends to focus on staying relevant more than being feature complete or even bug free. That would be tolerable, but MSFT takes the literal opposite approach to Apple who waits for a technology to mature enough that it's about to take off, tweaks it a little then releases a "magic" version and comes off looking like an innovative company. Instead, MSFT tends to look for the "next big thing" and gets on the bandwagon as soon as possible - usually long before the technology is ready for consumer products - and usually at 110%. Most of the time, it turns out that "next big thing" ends up being a niche product (Win 10 was 3D everything, 3D printing, modelling and "XR" for visualisation) or just wrong (Win 8 and "tablets are the future") forcing MSFT to backpedal after blowing tons of money. Win 11 was "The Web as the OS", AI (and weirdly "AI as the OS") and ARM/Copilot+ PCs, all of which have turned out to be mistakes or flops to some degree or entirely.
  • Recent Achievements

    • Reacting Well
      Gideon Waxfarb earned a badge
      Reacting Well
    • First Post
      NovaEdgeX earned a badge
      First Post
    • One Month Later
      pahariyaseo earned a badge
      One Month Later
    • Week One Done
      pahariyaseo earned a badge
      Week One Done
    • Week One Done
      hadiaali45 earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      427
    2. 2
      PsYcHoKiLLa
      144
    3. 3
      Nick H.
      90
    4. 4
      Steven P.
      81
    5. 5
      +Edouard
      80
  • Tell a friend

    Love Neowin? Tell a friend!