What redundancy exists at power level?
Here is the redundancy hierarchy at the power level:
Every physical server in an OCI data center (whether it’s a VM host or a Bare Metal machine) is equipped with redundant power supply units (PSUs).
A/B Power Feeds: These PSUs are connected to two completely independent power distribution paths within the data center. If one power cable or PDU (Power Distribution Unit) fails, the server continues to run on the second feed without a "reboot."
This is where you, the customer, have direct control. Each Availability Domain is split into three Fault Domains (FDs).
Hardware Isolation: Each FD is a group of hardware that does not share the same top-of-rack power source.
The Strategy: By placing Server A in FD-1 and Server B in FD-2, you ensure that a localized electrical failure (like a tripped breaker or a blown PDU) only affects 50% of your application.
In regions with multiple Availability Domains (ADs), each AD is a physically separate building.
Grid Independence: ADs are isolated from each other in terms of power and cooling. Often, they are connected to different electrical substations on the municipal grid.
Backup Systems: Every AD is equipped with its own:
UPS (Uninterruptible Power Supply): Battery arrays that provide 0-millisecond transition power if the grid blips.
Industrial Generators: On-site diesel or natural gas generators capable of running the entire data center for several days during a total grid blackout.
| Level | Redundancy Type | Protection Against... |
| Server | Dual PSUs (N+1 or 2N) | Single power cable/PSU failure |
| Rack (FD) | Independent PDUs | Rack-level electrical faults |
| Data Center (AD) | UPS & Generators | Municipal grid failure |
| Region | Geographic Separation | Catastrophic regional blackout |
By 2026, Oracle has integrated AI-driven power management into its data centers.
Real-time Load Shifting: If the system detects a potential thermal or electrical instability in one section of the data center, OCI can "live migrate" virtual machines to a different power zone (different Fault Domain) before a failure actually occurs.
Renewable Mix: OCI 2026 data centers are increasingly powered by local renewable microgrids (solar/wind + massive battery storage), providing a "tertiary" power source that is independent of the traditional utility grid.
Oracle's power redundancy isn't just for emergencies; it's for updates. When Oracle performs routine maintenance on data center electrical systems, they do so one Fault Domain at a time. Because your servers are spread across FDs, your "rented" infrastructure stays online while the physical power components "underneath" it are being serviced.
When looking at your Compute instance details in the console, check the Fault Domain field. If all your mission-critical servers say "Fault Domain 1," you have accidentally created a "power bottleneck." Move half of them to FD-2 or FD-3 to take full advantage of the power redundancy you're already paying for.