I'm an IT Manager who's been handling infrastructure procurement for about six years. In that time, I've personally made—and documented—47 significant mistakes, totaling roughly $10,300 in wasted budget. This article is the checklist I now use to prevent those errors.
It's tempting to think that buying a UPS is simple. You look at the wattage of your gear, you buy a box with a bigger number. But identical specs from different vendors at different price points can result in wildly different outcomes. The 'just add up the watts' advice ignores the terrible, non-linear reality of inrush current and transformer-based power supplies.
Here's a 6-step checklist I use for every UPS purchase now.
The 6-Step APC UPS Buying Checklist
Step 1: Stop Looking at Watts (Look at VA and Runtime First)
The biggest mistake I made in my first year (2017) was buying a 1500VA unit for a rack of three servers and a switch. The wattage was fine—well under the rated limit. But the runtime was a joke. We lost power for 15 minutes during a brownout, and the UPS died after 4 minutes. The servers didn't even have time to shut down gracefully.
Here's the deal: The VA (Volt-Amps) rating tells you the capacity of the UPS's internal transformer and battery inverter. The wattage rating tells you the max power draw it can sustain. But you need to check the runtime curve in the datasheet (note to self: always download the PDF, not just rely on the spec sheet on the sales page). A 1500VA APC Smart-UPS will give you about 15-20 minutes at 50% load, but only 5 minutes at 100% load. If you need to keep a critical server alive for a 10-minute generator startup, you need to oversize it.
Step 2: Profile Your Actual Power Draw (Don't Use Nameplate Ratings)
The second mistake? I used the PSU wattage listed on the back of the servers. A server with a 750W power supply running at idle might only draw 150W. But at startup? It can spike to 400W for a split second. If you add up the nameplate ratings, you'll buy a massive, expensive UPS you don't need. If you ignore the startup spikes, you'll buy one that crashes.
I now use a Kill-A-Watt meter (or a PDU with metering) to measure the actual load of a running server. Then I multiply that by 1.5 for headroom. For example, a server that draws 200W at idle gets a 300W allocation in my sizing. That was a hard lesson learned after I bought a 1000VA unit for a server that, at peak, surged past its limit during a firmware update (ugh).
Step 3: Understand the Type of UPS You Need (Standby, Line-Interactive, Online)
This is another area where the 'just buy a UPS' advice ignores nuance. An APC Back-UPS (Standby) is fine for a home office modem and router. For a server room with sensitive electronics? You need a Smart-UPS (Line-Interactive) or an Online/Double-Conversion unit.
I learned this the hard way in September 2022. We had a Back-UPS on a network switch. A voltage sag (brownout) happened, and the UPS switched to battery. But the transfer time (the gap between utility power failing and battery kicking in) was a few milliseconds too long for that switch's power supply. It rebooted. The whole office lost internet for 10 minutes (note to self: never use consumer gear for critical network infrastructure).
Per industry standards (and my own painful experience), here's the breakdown:
- Standby (Back-UPS): For basic devices with tolerant power supplies. Transfer time: ~4-8ms.
- Line-Interactive (Smart-UPS): For servers, switches, and gear with less tolerant PSUs. Transfer time: ~2-4ms. Also regulates voltage without using battery.
- Online (Double-Conversion): For absolute zero downtime. Power is always running through the inverter. Zero transfer time. Expensive but critical for MRI machines or high-end audio gear.
Step 4: Count Your Outlets (And Know the Difference Between Battery and Surge-Only)
This is a classic process gap. I once ordered a Smart-UPS 1500 for a rack with a server, a switch, a modem, a monitor, and a desk fan. The UPS had 8 outlets. I figured: plenty of room. But only 4 of those 8 outlets were backed up by the battery. The other 4 were surge-only. I had plugged my server and switch into surge-only outlets. The whole point of the UPS was defeated.
The third time I made this basic error on a $500 order, I finally created a label on the UPS that says, 'BATTERY: Ports A-D. SURGE: Ports E-H.' Should have done that after the first time.
Check the manual. The APC Smart-UPS 1500 manual (yes, you can find the PDF online) clearly maps which outlets are battery-backed. Don't trust the icons on the UPS itself—they can be ambiguous in a dark rack.
Step 5: Plan for the Network Management Card
Here's a step most people ignore. If you're buying a Smart-UPS for a business, you need the network management card (AP9630 or AP9631). This lets you monitor the UPS remotely, get email alerts, and perform graceful shutdowns of your servers.
I wasted a weekend in 2023 driving to the office because the UPS beeped and the office was empty. The beeping was a low-battery warning from a power outage that ended 3 hours before I got there. With a management card, I would have gotten an email at 2 AM and handled it remotely. Without it? I sat in traffic (ugh).
Factor in the cost of the card ($100-200 depending on generation) and the Ethernet cable. It's a no-brainer for a rack over $1,000.
Step 6: Don't Forget the Battery Replacement Cycle
APC UPS batteries last about 3-5 years. When you buy the UPS, set a calendar reminder for 36 months out to buy a replacement battery pack (RBC). The cost of a battery replacement is usually 30-50% of the cost of a new UPS. It's a deal-breaker for some budgets.
The fundamental lesson I learned after three battery failures? Don't wait for the UPS to tell you it's dead. By then, your servers have already crashed. Check the battery health via the management card quarterly.
Common Mistakes I've Seen
I've seen companies buy a UPS that exactly matches their current load, then complain when they add one more switch and the UPS can't handle it. Size for the future, not just the present. I made that mistake on a $3,200 order for a full rack setup in 2021.
Another one: buying the cheapest UPS from a generic brand to save $50. That UPS died after 18 months, and the replacement cost (plus the server crash it caused) was $2,000. The APC unit I bought instead cost $250 more but is still running 4 years later. The reliability premium is worth it.
Finally: don't mount a UPS at the bottom of a rack without a proper support shelf. UPS units are heavy. I saw a 1500VA unit fall from a rack shelf (it wasn't screwed in). It cracked the casing and damaged the battery terminals. The whole unit had to be replaced. That's a mistake that cost $700 in hardware plus cleanup time (note to self: always bolt the UPS to the rack rails).
If you follow this checklist, you'll avoid the dumb mistakes I made. I've caught 47 potential errors using this list in the past 18 months. It works.