1.3.1. Hardware Failure Patterns
💡 First Principle: Hardware failure follows the "bathtub curve"—high early failure rates (infant mortality from manufacturing defects), a long period of low failure rates, then increasing failures as components age. This is why warranty periods matter and why aging hardware needs proactive monitoring.
The most common hardware failure modes and their signatures:
| Component | Failure Mode | Symptoms |
|---|---|---|
| Hard drive (HDD) | Sector failures, bearing wear | SMART errors, slow read/write, clicking sounds, OS not found |
| SSD | Write endurance exhaustion | Performance degradation, write errors, sudden death |
| RAM | Cell failures, bit flips | Random crashes, BSODs, kernel panics, memory dump files |
| Power supply | Capacitor failure, overload | Random shutdowns, component damage, burning smell |
| CPU/GPU | Overheating | Thermal throttling, random reboots, system lockups |
| Fan | Bearing failure, debris | Rising temperatures, auditory indicators (grinding), thermal throttling |
| CMOS battery | Discharge | Incorrect date/time after reboot, BIOS settings reset |
| Backplane | Physical damage, connector wear | Multiple drive failures, intermittent drive connections |
Predictive failure analysis uses technologies like S.M.A.R.T. (Self-Monitoring, Analysis, and Reporting Technology) to detect early warning signs before catastrophic failure. Modern servers generate alerts when SMART thresholds are exceeded—this is the difference between scheduled maintenance and emergency recovery.
Hardware failures produce physical indicators: LED status lights on drives, NICs, and power supplies; LCD panel readouts on the chassis; POST codes during boot; and auditory cues (beep codes, grinding, clicking). Learning to read these indicators is a practical troubleshooting skill.
⚠️ Exam Trap: A POST failure doesn't necessarily mean CPU or RAM failure — POST codes identify which initialisation step failed, which could be any component on the bus (video, expansion cards, memory training). But be precise about where POST ends: a misconfigured boot order or a disconnected boot drive is not a POST failure. POST has already succeeded by then — the hardware initialised fine and the firmware simply cannot find something to boot. "No boot device found" is a boot failure; a beep code or a hung POST code is a POST failure. The distinction tells you whether to suspect hardware or configuration.
Reflection Question: Why is S.M.A.R.T. monitoring valuable even for drives that appear to be functioning normally?
