Data Center Power Explained
The Power Chain From Grid to Rack
Data center power is the chain of equipment that carries electricity from the utility grid to the servers in each rack, while protecting that supply from outages and poor power quality. Every link in the chain can fail, so facilities add redundancy at each stage, as explained in N+1 and 2N designs.
The grid supply usually arrives at high voltage and is stepped down by transformers on site. From there, switchgear (the protective switches and breakers that route and isolate power) feeds the UPS systems, and a distribution network carries power to the rows of racks. An automatic transfer switch (ATS) moves the load to backup generators when the grid fails, and back again when it returns.
Backup Power: UPS and Generators
- UPS (uninterruptible power supply): a battery-backed unit that conditions incoming power and supplies the load instantly when the grid drops. Its runtime is measured in minutes, long enough for generators to start and take over.
- Generators: engines, usually diesel, that start automatically during an outage and carry the load for as long as on-site fuel lasts. Facilities keep fuel storage and arrange deliveries to extend runtime during long events.
The two work together. The UPS bridges the gap, and the generator provides longer-term power. Batteries also protect equipment from short voltage dips and spikes that never become full outages. The energy lost along this chain is one reason efficiency is measured with power usage effectiveness (PUE).
Distribution, PDUs and Dual Feeds
A PDU (power distribution unit) takes a high-capacity feed and divides it into the outlets that serve each rack. Rack PDUs often include metering, which shows how much power each rack draws and helps operators plan capacity. How much power a rack can draw is covered in rack power density.
Most critical servers have two power supplies, and facilities run two separate paths, usually called A and B feeds. Each power supply in a server connects to a different path, so the server keeps running if one path fails. The protection only works if the two feeds are truly independent upstream. Two feeds that share one UPS or one switchboard are much less protected than they appear.
What Happens During a Grid Outage
Imagine the grid supply to a facility fails at night. The UPS batteries carry the load in the first moments, so servers see no interruption. The ATS detects the loss and transfers the load to generators, which have started automatically. When grid power returns, the transfer reverses and the UPS batteries recharge. In a well-designed facility, this whole sequence runs without manual action.
Frequently asked questions
What is a UPS in a data center?
A UPS (uninterruptible power supply) is a battery-backed unit placed between the power feed and the IT equipment. It supplies power immediately when the grid fails and smooths out voltage dips and spikes. Its runtime is usually short, measured in minutes, because it only has to bridge until generators take over.
Why do servers have two power supplies?
Each server with two power supplies connects one to feed A and one to feed B. If one path fails because of a UPS, breaker or cable fault, the server keeps running on the other. The benefit depends on the two paths being independent throughout the chain, not only at the rack.