How are firmware updates handled in rented servers?
Here is how the process works for the two different server types:
For standard VMs, OCI handles firmware updates almost entirely behind the scenes.
Live Migration: When a physical host needs a firmware update, OCI attempts to Live Migrate your running VM to a pre-patched host. You experience zero downtime; your server simply continues running on new hardware.
Reboot Migration: If your VM shape doesn't support live migration (e.g., Dense I/O with local NVMe), OCI schedules a Reboot Migration.
You are notified of a "Maintenance Window" (typically 14 days in advance).
You can proactively reboot the instance at a time that suits you.
When it reboots, it automatically starts up on a healthy, updated host.
On Bare Metal, you have direct access to the hardware, so firmware updates require more coordination.
Infrastructure Maintenance: You will receive a notification in the OCI Console under Compute > Instance Maintenance.
Rebuild in Place: Unlike VMs, a Bare Metal server often requires a "Rebuild in Place." During the scheduled window, the server is stopped, the firmware is flashed, and the server is restarted. This typically involves several hours of downtime.
The "Shift" Option: In 2026, many enterprise customers avoid the downtime by using OCI's Capacity Reservations. They spin up a new Bare Metal server, migrate their workload, and then terminate the old one before its maintenance window arrives.
In 2026, OCI offers Shielded Instances for both VMs and Bare Metal. This adds a layer of protection to the firmware update process:
Secure Boot: Ensures that only firmware signed by Oracle/Hardware vendors can boot.
Measured Boot: Uses a Trusted Platform Module (TPM) to "fingerprint" the firmware. If a firmware update were compromised or malicious, the server would refuse to boot, protecting your data from low-level "bootkit" attacks.
| Feature | Virtual Machines (VM) | Bare Metal (BM) |
| Typical Downtime | Zero (via Live Migration) | 2–4 Hours |
| User Action | None (Automatic) | Proactive Reboot or Wait |
| Notification | 14 Days (if reboot is needed) | 14 Days (Always) |
| Rollback | Handled by OCI | Handled by OCI during maintenance |
For the first time in 2026, OCI has opened up Advanced BIOS Settings for certain Bare Metal shapes.
Customer Control: While Oracle manages the version of the firmware for security, you can now toggle certain hardware features (like Hyper-Threading or SR-IOV) directly from the OCI Console without needing to enter the BIOS menu during a manual boot.
If a critical industry-wide vulnerability is discovered (like a new "Spectre" or "Meltdown" variant), Oracle reserves the right to perform Emergency Maintenance. In these rare cases, the 14-day notice may be shortened to 24–48 hours to ensure the entire cloud region remains secure.
Don't wait for an email. In 2026, you should check the Instance Maintenance dashboard in the OCI Console once a week. It shows you a "Due Date" for every server in your fleet. If you see a date approaching, use the "Reschedule" button to move the update to your weekend maintenance window.