I took over procurement for our company in 2020, and the question I get most often from other admins is some version of "what switch should we buy" or "do we need a network tester." And honestly, after five years and roughly 200 orders, my answer is always the same: it depends on who you are.
There is no single right network setup. What works for a 30-person office with no IT staff will be a mess for a 300-person company with a network team. So instead of giving you one answer, I'm going to break this into three scenarios. I'll tell you what I'd buy in each, what I'd skip, and how to figure out which one you're actually in.
Quick preview: Scenario A is the small office with no dedicated IT. Scenario B is the mid-size company with one or two IT people. Scenario C is the larger deployment with an actual IT team. If you already know which one sounds like you, skip ahead — but the judgment section at the end is where I'd start if you're not sure.
Scenario A: The 30-Person Office With No IT Staff
This was us until 2022. We had an office manager (me), a part-time contractor who came in when things broke, and about 30 people who all assumed the Wi-Fi was magic. I bought network gear the same way I bought office chairs — find something reliable, don't overthink it.
Here's the counterintuitive part: you probably don't need enterprise-grade routing or switching. Everything I'd read about business networking said to buy "future-proof" equipment. In practice, I found that half the features on a managed switch never got touched because nobody on staff knew how to configure them.
What actually worked for us:
- A straightforward managed switch — the Juniper EX2300 line is where I'd look now if I were redoing it. It's in the sweet spot of "manageable without a dedicated admin" but still reliable, and the PoE models cover phones and access points without a separate power injector, which saved us a ton of desk clutter.
- A basic network tester — not the $800 cable certifier, just a $60 link tester and a cheap cable checker. That was enough to figure out whether a run was dead or a port was flaky. Anything more sophisticated, we called the contractor.
What I'd skip: do not buy a high-density fixed switch like the EX5120 Juniper switch for an office this size. It's overkill, and — this is important — a lot of EX5120 units floating around on secondary markets are near or past end-of-life. This was true ten years ago when the EX5120 was a current workhorse and buying used made sense. Today, the lifecycle math has changed and firmware support is the thing that bites you. I got burned on this exact pattern with a different vendor's gear. The invoice looked great until the security patches stopped. (Note to self: always check the end-of-support date before signing anything.)
On the office device management side: at this scale, you're probably not running an MDM. That means "how to reset phone when locked" becomes a real, recurring headache. We had employees showing up with locked personal phones tied to work accounts, and there was no clean process. If I could do it over, I'd standardize on a cheap MDM tier just to have a reset path that doesn't involve the employee's personal Apple ID — which, honestly, is a conversation nobody wants to have on a Monday morning.
Scenario B: The 100–300-Person Company With 1–2 IT People
This is where the math shifts. Once you have someone whose actual job includes network management, the "set it and forget it" approach starts costing more than it saves. Your IT person ends up spending half their week on manual troubleshooting that a better platform would handle.
This is where Juniper's Mist AI starts to matter, and I want to explain it in procurement terms rather than marketing terms. The value isn't "AI" as a buzzword — it's that the system tells you what's wrong before your users do. For a 150-person company with one IT admin, that's the difference between a proactive Monday morning and a fire drill.
What I'd budget for in this scenario:
- Juniper EX2300 switches at the access layer, with PoE models for wireless access points. We compared quotes against a couple of alternatives in 2024, and for our port count and PoE budget, the EX2300 hit the right balance. Prices varied by roughly 30% across resellers for identical specs — worth getting at least three quotes.
- A real network tester at this point — something that can do cable certification and link testing. Our IT admin used a Fluke Linksprinter, and it paid for itself just in avoided callbacks to cabling contractors.
- A proper MDM. At this size, "how to reset phone when locked" shouldn't eat up an IT ticket every week. Self-service reset sounds trivial until you're managing 200 devices.
The causation thing that took me a while to understand: people think expensive network gear delivers better uptime. Actually, it's the other way around — gear that delivers consistent uptime gets to charge more, because the vendor knows the total cost of downtime. Once I reframed it that way, I stopped shopping purely on sticker price and started asking vendors about support response times and lifecycle commitments. That question alone filtered out half the resellers we talked to.
Scenario C: The 300+ Employee Deployment With a Real IT Team
If you're at this scale, you're not reading a procurement blog for the switch recommendation — you've got network engineers. But from the purchasing side, there are still things I'd insist on, mostly around total cost of ownership.
At this size, push for a conversation about automation. Juniper's Mist platform and its AI-driven network management isn't a nice-to-have when you're managing dozens of sites or hundreds of access points — it's what lets a small IT team keep up. The Juniper EX2300 and higher-end QFX series both fit here, depending on whether you're buying access or aggregation gear.
About the EX5120: if a reseller is pitching you a great deal on these, ask two questions. First, what's the end-of-support date? Second, what's the migration path? The EX5120 was a workhorse in its day, but "in its day" is the key phrase. For new deployments in 2025, I'd be looking at current-generation gear and treating EX5120 units as a specific-fit or lab scenario, not a primary production purchase. So glad I asked that question on a recent quote — almost signed for a batch that would have been out of support by year two.
At this scale, a network tester isn't a single device — it's part of your tooling budget. Cable certifiers, Wi-Fi analyzers, link runners. Let your network team spec it, and just make sure the purchase order goes through you so someone is tracking renewals and calibration dates.
One more thing at this level: the device lifecycle conversation gets formal. You'll have a documented refresh cycle, a disposal process, and — yes — an actual answer for "how to reset phone when locked" that doesn't involve a help desk ticket. That's the sign you've graduated out of the previous scenario.
How to Tell Which Scenario You're In
Ask yourself three questions:
- Does anyone on staff have "network" in their job description? If no, Scenario A. If yes, and it's one person, Scenario B. If you have a team, Scenario C.
- How many locations are you supporting? One office, Scenario A or B. Multiple sites or a data center, Scenario C.
- What's your annual network equipment spend? As a rough rule of thumb I use: under $10K, don't over-buy. $10K–$100K, you're in the automation zone. Above that, you should have a dedicated owner and a lifecycle plan.
The most useful piece of procurement advice I can give: match the equipment to the capability you currently have, not the capability you hope to build. I've seen companies buy enterprise gear they never configured, and I've seen companies outgrow consumer equipment in 18 months. The sweet spot is to buy one step ahead of where you are — not three.
And beware the vendor who says they can do everything. The ones who've told me "we don't do that well, you should talk to X" are the ones I still work with. That's the boundary test, honestly. A supplier who knows where their expertise ends is easier to trust everywhere else. Speed, price, and support — pick two, and make the supplier tell you which two they're actually good at.