PDU / Rack Calculator
Number of servers per rack based on power capacity (watts).
Rack power and power per server
Table of Contents
Comprehensive PDU and Rack Power Guide
What is PDU and rack power?
A PDU (Power Distribution Unit) is a device that distributes electrical power to servers, switches, and other equipment inside a rack. The rack's total power capacity is limited by the PDU rating, the circuit breaker feeding it, and the capacity of the datacenter feed. Planning how many servers can be powered from one rack—based on their power draw in watts—is essential to avoid overloads, tripped breakers, and unsafe operation.
How rack power planning works
Rack power planning involves knowing the available capacity (PDU or circuit limit), the power draw per device, and applying a safety margin. The key steps are:
- Available capacity: The PDU rating or circuit breaker limit (e.g. 3 kW, 6 kW, 10 kW per rack or per PDU).
- Power per server: Nameplate or measured power per server (or average if mixed).
- Safety margin: Use only a percentage of capacity (e.g. 80%) for continuous load to leave headroom for peaks.
- Calculation: Usable power = Capacity × (Margin % / 100). Max servers = Usable power ÷ Power per server (rounded down).
- PDU: Power Distribution Unit; may be basic (outlets only) or metered/switched. Rating in amps or watts (e.g. 30 A × 120 V ≈ 3.6 kW).
- Circuit breaker: Protects the circuit; the rack cannot draw more than the breaker rating without tripping.
- 80% rule: Continuous load should not exceed 80% of circuit/PDU rating to allow for peaks and avoid nuisance trips.
- Dual-cord: Servers with two power supplies plug into two different PDUs; each PDU must have capacity for half the servers (or full if one PDU fails).
- Nameplate vs measured: Nameplate is maximum; measured under real load is often lower. Use the higher for planning to be safe.
Single PDU vs redundant power
Single PDU per rack
- One power path; cost-effective
- PDU or circuit failure = entire rack down
- Plan capacity = PDU rating × margin
- Common in non-critical or dev environments
Redundant (A+B) power
- Two PDUs per rack; dual-corded servers
- One PDU can fail; servers stay on
- Each PDU must support full rack load (N+1) or half (2N)
- Standard for production and high availability
Benefits of power planning
- Avoid overloads: Prevents tripping breakers and unexpected outages when all servers are under load.
- Predictable density: You know how many servers fit per rack before deployment.
- Compliance and safety: Many standards and insurers expect documented power limits and margins.
- Capacity planning: When adding servers, you know if you need a new circuit or PDU.
Limitations and considerations
- Power per server varies with workload; use nameplate or peak measured for conservative planning.
- Network switches, storage, and other gear in the same rack share the same PDU—subtract their power from available capacity.
- Voltage and phase (single vs three-phase) affect actual capacity; the calculator uses watts for simplicity.
- In redundant setups, plan per PDU: each path must support the load that will run on it during a failure.
Do not exceed the PDU or circuit capacity. Continuous load above the rating can cause overheating, breaker trips, or fire risk. Use the 80% rule (or your site's standard) and plan for the worst-case power draw of your equipment.
Power density trends
Modern datacenters are moving toward higher power per rack (e.g. 10–20 kW and more for high-density compute). PDUs and circuits must be sized accordingly. This calculator works for any capacity: enter your rack or PDU limit in watts or kW, and the power per server, to get the maximum number of servers that fit within your margin.
Choosing rack and PDU capacity
When designing or refitting a rack, consider:
- Server mix: Average power per server (or use the highest if you want a single number for the whole rack).
- Redundancy: Single or dual power path; if dual, each PDU must support the load that remains if the other fails.
- Future growth: Leave headroom or plan for a second circuit/PDU when you add more servers.
- Site limits: The datacenter may cap power per rack; never exceed that limit.
Typical power values
See the reference table in section 6 for typical power ranges per server type and common PDU ratings. Use these as a starting point when you do not have nameplate or measured values.
Planning rack power with a PDU/rack calculator ensures you never overload a circuit and that you know how many servers can safely fit in a rack. Use the 80% rule (or your site standard), account for non-server loads, and in redundant designs plan capacity per PDU. This tool gives you the maximum number of servers for a given capacity and power per server; combine it with nameplate or measured data for accurate results.
Rack Power Overview
Rack power is limited by the PDU and the circuit feeding it. To avoid overloads, plan the number of servers based on their power draw and a safety margin (e.g. 80% of capacity).
- Usable power = Rack capacity × (Margin % / 100)
- Max servers = Usable power ÷ Power per server (rounded down)
- 80% margin is the common standard for continuous load
Rack / PDU power capacity
The rack or PDU power capacity is the maximum power (in watts or kilowatts) that can be drawn from that PDU or circuit. It is determined by the PDU's rating and the circuit breaker. Common values are 3 kW, 6 kW, 8 kW, or 10 kW per PDU; high-density racks may have 20 kW or more. Enter this value in the calculator as the total available for the rack (or per PDU if you have two and plan for one path).
- Defined by PDU nameplate and/or circuit breaker
- Often 3–10 kW per PDU in standard racks
- Enter in W or kW in the calculator
Power per server
Power per server is the electrical draw of one server, usually in watts. Use the nameplate (maximum) value for conservative planning, or the measured value under typical or peak load. Mixed racks: use the highest draw or an average weighted by count. Typical ranges: 1U 150–400 W, 2U 300–500 W, blades 200–400 W per blade.
- Nameplate = max; measured = real load
- Use worst-case for safe planning
- Varies by form factor and workload
Utilization and safety margin
The utilization or safety margin is the percentage of the PDU/circuit capacity that you allow for continuous load. Using only 80% (default in the calculator) leaves 20% headroom for power spikes, inrush, and measurement error, and helps avoid nuisance breaker trips. Some sites use 70% or 85%; enter the value that matches your standard.
- 80% is the common standard (80% rule)
- Reduces risk of breaker trips
- Configurable in the calculator
Typical power values
| Equipment | Typical power (W) |
|---|---|
| 1U server (light) | 150 – 250 |
| 1U server (standard) | 250 – 400 |
| 2U server | 300 – 500 |
| Blade (per blade) | 200 – 400 |
| Network switch (1U) | 50 – 150 |
| Rack PDU (common) | 3 – 8 kW per PDU |
Best practices
- Use nameplate or measured power per server; idle power is lower than under load.
- Leave headroom (e.g. 80% rule) to avoid breaker trips during power spikes.
- Account for network gear, storage, and other devices in the same rack if they share the same PDU.
- Dual-corded servers: each cord feeds from a different PDU; plan capacity per PDU, not per server.
- Document your power plan and margin; review when adding or replacing equipment.