Power distribution in data centers

Power is the constraint a data center is designed around. Understanding the chain from the street to the server explains most of the rules technicians are asked to follow, and all of the ones that are non-negotiable.

The chain

1. 1Utility feed enters the site, often from more than one substation.
2. 2Generators stand by to carry the site if utility fails, typically starting within seconds.
3. 3UPS systems bridge the gap between utility loss and generator load acceptance.
4. 4Switchgear distributes to the floor, splitting into independent A and B paths.
5. 5Remote power panels feed the rows, and breakers feed individual racks.
6. 6Rack PDUs deliver outlets, and the server's dual power supplies take one from each side.

Why A and B exist

Every element from the UPS down is duplicated so that any single component can be taken out of service without dropping load. That only holds if equipment is actually connected to both sides. A dual-corded server with both cords on the A feed is a single point of failure wearing a redundancy costume.

Breakers and headroom

Circuits are rated for continuous load at 80% of the breaker rating. A 20A circuit therefore supports 16A continuously. Before adding equipment, check the nameplate draw against measured PDU load and confirm the new total stays under that ceiling on both sides — during normal operation and in the failure case where one side carries everything.

What technicians own

• Connecting equipment to the correct A and B outlets, and recording which ones.
• Reading and reporting PDU load, and flagging imbalance before it becomes a risk.
• Responding to power alarms with the correct escalation path rather than resetting breakers.
• Following the Method of Procedure exactly during maintenance windows.
• Never assuming a circuit is dead. Verify, lock out, tag out.

Electrical work beyond the outlet belongs to critical facilities engineers. Knowing precisely where your responsibility ends is itself a competency interviewers test for.