How to Choose a UPS for a Small Server Rack

The short answer: Add up what your equipment actually draws in watts, size the UPS so that load sits at no more than about 80% of its watt rating, and choose a line-interactive or online (double-conversion) UPS with pure sine wave output. Then make sure it can shut your servers down cleanly before the batteries run out.

Watts matter more than VA

UPS ratings show two numbers: volt-amps (VA) and watts (W). The watt rating is the one that matters for servers. A UPS labelled 1500VA might only support 1000W, depending on its power factor, so size to the watts figure.

Step 1: work out your real load

1. List everything that will plug into the UPS: servers, storage, switches, the firewall and anything else in the rack.
2. Find each device's typical power draw. A power supply's rating (for example 800W) is its maximum, not what the server uses. Measured figures from iDRAC, iLO or a metered PDU are best. Otherwise use the vendor's calculator: Dell's Enterprise Infrastructure Planning Tool and HPE Power Advisor both estimate draw for a given configuration.
3. Add the figures together.

Step 2: add headroom

Don't size a UPS to run at 100%. Aim for your load to sit at no more than about 75–80% of the UPS's watt rating. That leaves room for growth and start-up surges, and it gives you more runtime.

Example: two servers drawing 350W each, a NAS at 100W and a switch at 50W come to 850W. At an 80% target you need a UPS rated for at least 1,063W, so a model rated around 1,500W gives comfortable headroom.

Step 3: choose the right type

• Standby (offline): fine for a desktop PC, but not suitable for servers.
• Line-interactive: corrects small voltage changes without switching to battery. A good fit for most small server racks, as long as it has pure sine wave output.
• Online (double-conversion): equipment always runs from the inverter, so it gets clean power with no transfer time. Choose this for critical systems, or where the mains supply is unreliable.

Most modern server power supplies use active power factor correction and expect a pure sine wave. Simulated or stepped sine wave output can make them shut down when the UPS switches to battery.

Step 4: decide how much runtime you need

For most businesses, a UPS isn't there to keep servers running for hours. Its job is to ride through short cuts and give systems enough time to shut down cleanly, typically 10 to 15 minutes. With a generator, it only needs to bridge the gap until the generator takes the load. If you need longer, look for models that take external battery packs. Runtime falls as load rises, so check the manufacturer's runtime chart at your actual load.

Step 5: plan the automatic shutdown

Runtime only helps if something uses it. Fit a network management card, or connect the UPS by USB, and install the manufacturer's shutdown software, such as APC PowerChute or Eaton Intelligent Power Manager. It shuts down virtual machines and hosts in the right order before the batteries run flat. Test it before you need it.

Step 6: check the practical details

• Form factor: rack-mount units take 2U to 3U or more and are heavy. Check your rack's depth and rails.
• The supply to the UPS: a standard UK 13A plug can supply up to about 3kVA. Larger units need a dedicated 16A or 32A commando socket fitted by an electrician.
• Outlets: count the IEC C13 and C19 sockets against your equipment's power cords.
• Batteries: sealed lead-acid batteries typically need replacing every three to five years. Lithium-ion units cost more but last longer.

Need help?

If you're planning a new rack or replacing an ageing UPS, request a quote and our team will size it against your equipment.