Is hardware replacement automatic?
In 2026, the answer is a resounding yes for Virtual Machines, while Bare Metal servers require a slightly different approach. OCI has moved toward a "self-healing" infrastructure model where hardware issues are often resolved before you even notice them.
Here is how hardware replacement is handled for your rented servers:
For standard VM instances, OCI uses a technology called Live Migration to handle failing hardware.
Proactive Detection: OCI’s monitoring system constantly checks for "pre-failure" signals (like high memory error rates or fluctuating power levels on a physical host).
Automatic Migration: If a physical server is deemed "unhealthy," OCI automatically moves your running VM to a healthy host in the background.
Zero Downtime: Your application keeps running, your IP address stays the same, and your data remains intact. This is the definition of "automatic hardware replacement."
Because you are renting the entire physical machine with no virtualization layer, OCI cannot "live migrate" a Bare Metal instance.
The Notification: If OCI detects a hardware issue on your physical server, you will receive an Infrastructure Maintenance notification.
The Process: You are typically given a 14-day window to "Reboot Migrate" your instance. When you trigger the reboot, OCI automatically provisions your environment on a brand-new physical server and attaches your existing storage.
Auto-Recovery: If you don't act before the deadline, OCI will eventually perform a forced reboot migration to a healthy server to prevent a catastrophic failure.
The disks "underneath" your servers are handled even more aggressively.
Distributed Parity: Your data is broken into pieces and spread across many physical disks.
Automated Reconstruction: If a physical drive fails in the data center, the OCI storage fabric automatically reconstructs the missing data on a spare drive. As a customer, you never see a "Disk Replacement" ticket; the system simply heals itself in the background.
| Instance Type | Replacement Method | User Impact |
| Virtual Machine | Live Migration | None (System stays up) |
| Bare Metal | Reboot Migration | Brief reboot to new hardware |
| Storage (Volumes) | Distributed Healing | None |
| GPU Instances | Reboot Migration | Brief reboot to new hardware |
In 2026, you can tell OCI exactly how to behave when hardware fails via the Availability Configuration menu:
Use Live Migration if possible: (Default) Best for uptime.
Reboot Migrate: Best if your application is extremely sensitive to the slight CPU "jitter" that can occur during a live migration.
Opt-out: You handle the migration yourself manually within a set window.
This is the one exception to automatic replacement. If you are using Dense I/O shapes with local NVMe SSDs (the "super fast" internal drives), that data is physically tied to the server.
The Risk: If that hardware fails, the data on the local drive is lost.
The Solution: In 2026, OCI recommends using Volume Groups or Software RAID to mirror that local data to a second instance. Always remember: OCI replaces the server automatically, but it can only "save" the data that is stored on Block Volumes, not local temporary drives.