BIOS/UEFI Configuration and Out-of-Band Management

BIOS/UEFI Configuration and Out-of-Band Management

What BIOS and UEFI Do

Every server contains firmware that runs before the operating system loads. The older standard is BIOS (Basic Input/Output System); the modern replacement is UEFI (Unified Extensible Firmware Interface). UEFI supports drives larger than 2 TB, faster boot times, Secure Boot, and a graphical setup interface. In a data center you will encounter both, but new servers ship with UEFI by default. Regardless of which firmware a server uses, it is your entry point for hardware-level configuration that the OS cannot change.

Key Settings You Will Configure

Boot device priority tells the server which device to try first when starting: typically the primary SSD or hard drive array, followed by the network (PXE boot), then optical or USB. Set PXE boot last unless you are deliberately deploying an OS over the network, because an accidental network boot can stall a production server.

UEFI vs. Legacy (CSM) mode controls how the firmware hands off to the OS loader. Keep servers in UEFI mode unless the OS or hardware explicitly requires legacy mode; Secure Boot only works in UEFI mode.

Secure Boot validates that the OS bootloader is signed by a trusted authority before executing it. Enable Secure Boot on servers running supported OS versions to protect against bootkits and unauthorized code at startup.

Memory operating speed (DRAM frequency) is usually set to "Auto," which lets the firmware negotiate the highest speed the installed DIMMs and CPU memory controller support. A spec sheet may require you to set a specific speed for workload predictability or hardware qualification. Never set a speed higher than the DIMM's rated speed.

Unused onboard ports (unused SATA controllers, serial ports, embedded video when a PCIe GPU is present) should be disabled to reduce attack surface and free interrupt resources. Always document what you disable so the next technician knows the state.

Out-of-Band Management: IPMI, iDRAC, and iLO

Out-of-band (OOB) management lets you access a server's power controls, hardware sensors, and a virtual console even when the main OS is crashed, the server is powered off, or you are physically in a different building. It runs on a dedicated management controller with its own processor, memory, and network port—completely separate from the server's main CPUs and data NICs.

• IPMI (Intelligent Platform Management Interface) is the open industry standard protocol underlying most OOB systems. It defines how management software communicates with the Baseboard Management Controller (BMC).
• iDRAC (Integrated Dell Remote Access Controller) is Dell's implementation. It provides a web interface, a command-line via SSH, and a virtual console (remote KVM).
• iLO (Integrated Lights-Out) is HPE's equivalent product with similar capabilities.

For OOB management to work, the management port must have a valid IP address, subnet mask, and default gateway on a dedicated management VLAN—separate from production traffic. You configure this address either inside the BIOS/UEFI setup screens or through the BMC's own setup utility (often reached during POST by pressing a key combination shown on screen).

Navigating the Web Interface

Once the management IP is set and your workstation is on the same management network, open a browser, type the IP address, and log in with the administrator credentials. Inside the iDRAC or iLO interface you will find:

• Virtual Console — a real-time view of what is on the server's screen, with keyboard and mouse input. Use this to complete OS installation or troubleshoot a crash without being physically present.
• Power Control — buttons to gracefully shut down, force power off, power on, or perform a cold reset. A graceful shutdown sends the signal to the OS. A force power off cuts power immediately—equivalent to holding the physical power button.
• System Event Log (SEL) — a timestamped record of hardware events: fan failures, temperature thresholds, memory errors, power supply faults. Review the SEL before and after any hardware change.

Remote Power Cycle Procedure

A remote power cycle is a routine task for a data center technician. The correct sequence is: (1) verify you are connected to the correct server by checking the service tag displayed in the management interface against the physical label; (2) attempt a graceful shutdown first unless the server is unresponsive; (3) confirm the physical power LEDs show the server is off before initiating power-on; (4) document the time, interface used, and reason in your configuration record.

Safety Considerations

Out-of-band management gives you the ability to cut power to a running server from across the room or across the country. Always confirm server identity before issuing any power command—a mistaken force-off on the wrong server can cause unplanned downtime and data loss. When physically working inside a server, follow lockout/tagout (LOTO) procedures: power down through the OS, then cut power at the PDU breaker and apply your lock before opening the chassis. OOB management does not eliminate the need for LOTO—the BMC still has standby power even when the server appears off. Wear an ESD wrist strap or use an ESD mat when handling circuit boards and DIMMs. Do not modify BIOS settings on a production server without a change-management approval and a rollback plan.