What is generator backup planning for racks?
If your UPS is the "sprint runner" that keeps your servers alive for 15 minutes, the Generator is the "marathoner" that keeps them alive for 15 days. In a professional hardware environment, generator planning is the difference between a minor blip and a catastrophic data center outage.
When the utility power goes dark, a complex sequence of mechanical and electrical events must occur perfectly to keep your racks humming.
A generator doesn’t start instantly. When the power fails, the generator needs anywhere from 10 to 30 seconds to crank the engine, reach the correct RPM, and stabilize the voltage.
The Automatic Transfer Switch (ATS) is the brain of this operation. It monitors the utility line; the second it detects a failure, it signals the generator to start. Once the generator is "warm," the ATS flips the load from the street to the engine.
The Hardware Link: During those 30 seconds, your UPS (Uninterruptible Power Supply) is the only thing keeping your servers from crashing. Your generator planning must ensure that your UPS battery runtime is at least 3x longer than your generator's start-up time.
You can’t just add up the wattage of your servers to size a generator. When a data center full of servers, cooling units, and pumps all try to pull power simultaneously after a switchover, they create a massive "spike" known as Inrush Current.
Step Loading: Professional planning involves "stepping" the load. You don't dump the whole data center on the generator at once.
Step 1: Critical Servers and UPS charging.
Step 2: Primary Cooling (Chillers/Fans).
Step 3: Lighting and non-critical systems.
The Rule of Thumb: Most engineers size generators to 1.5x or 2x the actual "steady-state" load of the racks to handle these spikes without the generator stalling.
A generator is a mechanical engine, and like a car, it requires maintenance to ensure it starts in a crisis.
Fuel Polishing: Diesel fuel can degrade over time, growing algae or collecting water. Backup plans must include "fuel polishing" (filtering) to ensure the engine doesn't clog during a blackout.
Block Heaters: In cold climates, generators use electric heaters to keep the engine block warm 24/7. This ensures that when the power fails, the engine can reach full speed in seconds rather than minutes.
Runtime Planning: Standard planning requires at least 48 to 72 hours of fuel on-site. For Oracle Cloud or Tier IV facilities, this can extend to a week with priority refueling contracts in place.
Generators produce "dirty" power compared to the utility grid. The frequency might wobble slightly (59Hz instead of 60Hz).
Some sensitive UPS systems will refuse to "accept" generator power if it’s too noisy, staying on battery until they die. When planning, you must ensure your UPS is configured for "Generator Mode," which allows for a wider tolerance of frequency fluctuations.
Generator planning isn't just about buying a big engine; it’s about timing. It’s a choreographed handoff between the utility, the UPS, and the diesel engine. If the handoff is seamless, your Oracle database never knows the city has gone dark. If it fails, your hardware is just an expensive collection of bricks.