How does dual power feed architecture work?
In the world of high-availability hardware, there is one absolute rule: Everything fails. Cables fray, circuit breakers trip, and Uninterruptible Power Supplies (UPS) eventually need maintenance.
To prevent these inevitable glitches from crashing your data center, we use Dual Power Feed Architecture (often called 2N redundancy). This design ensures that even if an entire side of the power grid goes dark, your servers never miss a beat.
A dual power feed isn't just about having two plugs; it’s about having two entirely independent "lives" for your equipment. Here is how the path looks from the utility to the server:
Independent Sources (A & B): Ideally, "Feed A" and "Feed B" come from different utility substations or, at the very least, separate main switchboards.
Separate UPS Systems: Feed A goes to UPS A, and Feed B goes to UPS B. This ensures that if one battery system fails or catches fire, the other remains operational.
Independent PDUs: As we discussed in our
The "magic" happens inside the server's chassis. Enterprise-grade hardware (like Oracle Sun Servers or Cisco UCS) comes with Dual (Redundant) Power Supplies (PSUs).
Load Sharing: Under normal conditions, both PSUs are active. They split the work, each pulling 50% of the required power.
The Failover: If PDU A loses power, PSU B instantly senses the drop and takes over 100% of the load. This happens in milliseconds—faster than the server’s internal capacitors can drain—meaning the OS never even realizes there was a power hit.
When designing your power architecture, you'll often hear these two terms. It’s important to know the difference:
N+1 (Parallel Redundancy): You have "N" (the number of units needed) plus one extra for backup. If you need 3 UPS units, you buy 4. This protects against a single unit failing but not against a total path failure (like a main breaker tripping).
2N (System Redundancy): You have two complete, independent systems. If System A dies entirely, System B can carry the full load. This is the gold standard for Tier III and Tier IV data centers.
Even with a dual feed, human error can create a single point of failure:
The "Single-Corded" Device: Some hardware (like small switches or older firewalls) only has one power cord. If you plug that into PDU A, and PDU A fails, that device dies—potentially taking down the whole network.
Solution: Use an ATS (Automatic Transfer Switch), a small device that takes two power inputs and provides one output to the single-corded device.
Overloading during Failover: If both sides of your 2N system are running at 60% capacity, and one side fails, the remaining side will try to pull 120%. This will instantly trip the remaining breaker, causing a total rack collapse.
Rule of Thumb: Never let a single PDU in a dual-feed setup exceed 40-45% utilization.
Dual Power Feed architecture is the ultimate insurance policy for your hardware. By eliminating every single point of failure between the power plant and the processor, you ensure that your "Always On" services actually stay on.