I can’t count how many times I’ve arrived on site to install a UPS that looked perfect on paper. The client bought the 1500 VA model because the server only draws 500 watts, the box claimed nearly an hour of runtime, and everyone assumed the problem was solved.
Then the power goes out.
Instead of an hour, the UPS lasts twelve or fifteen minutes. Sometimes less. Nothing is broken. The marketing just wasn’t describing the same conditions as the real world.
VA and watts are not the same thing
The first mistake usually happens before anyone even looks at runtime charts.
A UPS is typically advertised in volt-amps (VA). Your equipment consumes watts. Those numbers are related, but they are not interchangeable.
The difference comes down to power factor. Modern server power supplies are generally efficient enough that the gap isn’t huge, but it still matters. A 1500 VA UPS is not automatically capable of delivering 1500 watts.
Every manufacturer publishes both ratings. Ignore the VA figure for capacity planning and look at the maximum watt rating instead. If your equipment exceeds that number, the UPS is undersized regardless of what the VA rating says.
Measure the load you actually have
I’ve seen people estimate server power based on the size of the power supply more times than I can remember.
A server with dual 800-watt power supplies is not drawing 1,600 watts. Those are maximum capacities, not normal operating loads.
Measure the equipment instead. A managed PDU, a smart UPS, an inline power meter, or even intelligent outlet monitoring will tell you far more than the label on the back of the chassis. Virtualisation hosts often idle surprisingly low. File servers packed with spinning disks can be much higher than expected. Network switches barely change under load.
Real measurements beat assumptions every time.
A quick example
Imagine a virtualisation host that draws 420 watts under its normal workload. You find a UPS rated for 1500 VA and 900 watts.
On paper, that looks oversized. In reality, the server is already using about 47% of the UPS’s watt capacity. Looking only at the “1500 VA” label makes it feel like there’s loads of room left, but the runtime chart tells the real story. At roughly 420 watts, you might only get 20 to 30 minutes depending on the model and the age of the batteries.
Now add a storage array that consumes another 120 watts. Your total load jumps to 540 watts, or about 60% of the UPS’s rated capacity. You haven’t exceeded its limits, but your available runtime drops noticeably. Nothing changed except another device was plugged in.
This is exactly why I always work from measured wattage and the manufacturer’s runtime chart instead of the headline number on the box.
Runtime is not linear
This is where most expectations fall apart.
People naturally assume that if a UPS runs for 20 minutes at full load, it should run for 40 minutes at half load. Battery systems don’t work that way.
As the load increases, available runtime drops faster than most people expect. The discharge curve isn’t linear, and manufacturers know this. That’s why the detailed runtime charts buried in the documentation rarely resemble the number printed on the front of the box. That headline runtime often represents a very light load that doesn’t match production equipment.
Always size against the runtime chart for your expected wattage, not the marketing claim.
Why I like the 40% rule
When I’m sizing a new installation, I generally aim for the normal operating load to land somewhere around 40% of the UPS’s rated watt capacity. It’s not a magic number. It’s a practical one.
At around that range you typically get several benefits:
- Better runtime during an outage
- Capacity for future equipment
- Lower battery stress
- Less chance of unexpectedly exceeding capacity during startup or peak loads
Running a UPS at 80 or 90% all day usually works until something changes. Another server gets added. Storage expands. A switch stack grows. Suddenly there’s no headroom left. Buying one size larger at the beginning is often cheaper than replacing the entire unit a year later.
Batteries get older even if nothing else changes
A five-year-old UPS isn’t the same UPS you bought.
Lead-acid batteries lose capacity over time. Temperature accelerates the process. Poor ventilation makes it worse. The UPS may still pass its self-test while providing significantly less runtime than it did when it was new.
I’ve seen installations where everything appeared healthy until the first real outage. The batteries were simply no longer capable of supporting the expected load. If runtime is important to your business, battery replacement shouldn’t be an afterthought. It should be scheduled maintenance.
Planning is easier than guessing
UPS sizing isn’t difficult once you have the right inputs. You need the actual wattage of the equipment, the runtime you want, and realistic expectations about battery ageing and future growth. Everything else is just arithmetic.
If you don’t want to work through the calculations manually, there’s a free UPS Runtime Calculator in the MB Networks SysAdmin Toolbelt. It estimates runtime using real equipment loads and makes it much easier to compare different UPS sizes before buying hardware.
Buy for the outage you expect
The goal isn’t to keep the office running for four hours. Most businesses need enough runtime to ride through short outages, allow generators to start, or give servers time to shut down cleanly. Those are very different requirements.
Figure out how long your systems actually need to stay online. Measure your real power consumption instead of estimating it. Then size the UPS around those numbers, leaving enough headroom that the next hardware upgrade doesn’t force you to replace it.
Five minutes with a power meter and a runtime calculator will give you a far more accurate answer than anything printed on the front of the box.
Questions or corrections? info@mb-networks.ca