2.1.2. Server Components and Compatibility
💡 First Principle: Every component you install into a server must be on the Hardware Compatibility List (HCL) for that server model. The HCL isn't a suggestion—non-HCL components risk driver conflicts, firmware incompatibility, voided warranties, and unpredictable behavior under load.
Core Server Components
| Component | Function | Server-Specific Considerations |
|---|---|---|
| CPU | Processes instructions; server CPUs (Xeon, EPYC) support more cores, larger cache, multi-socket | Check socket type, TDP (thermal design power), core count for workload |
| GPU | Graphics processing; in servers, used for compute acceleration (ML, video rendering) | Requires adequate PCIe bandwidth and power delivery |
| RAM | Temporary working memory; servers use ECC | Match speed, generation (DDR4/DDR5), and population rules (DIMMs in pairs/quads) |
| Expansion cards | Add capabilities (NICs, HBAs, RAID controllers, GPU) via PCIe slots | Check PCIe generation (Gen 3/4/5) and slot width (x1, x4, x8, x16) |
Bus and Interface Types
Understanding bus types is important for both installation and troubleshooting:
- PCIe (Peripheral Component Interconnect Express): The dominant expansion bus. Lanes (x1, x4, x8, x16) determine bandwidth—a 10 GbE NIC typically uses PCIe x4 or x8.
- PCI-X: Older parallel bus, largely replaced by PCIe. Still appears in legacy hardware questions.
- SATA (Serial ATA): Internal storage interface for hard drives and SSDs. Common in workstations and entry-level servers.
- SAS (Serial Attached SCSI): Server-grade storage interface—faster, more reliable, supports longer cable runs and more devices than SATA. SAS controllers are backward-compatible with SATA drives; SATA controllers cannot use SAS drives.
- NVMe (Non-Volatile Memory Express): Ultra-fast SSD interface over PCIe, bypassing SATA limitations. Latency ~10–20 µs, against ~50–100 µs for a SATA SSD — both are microseconds. It is the spinning disk that works in milliseconds (~5–20 ms), which is the two-orders-of-magnitude gap that matters.
The HCL Process
Before purchasing any component for a server, check the manufacturer's HCL:
- Identify the server model and current firmware version
- Look up the component (drive, NIC, RAM, expansion card) on the vendor's HCL
- Confirm the specific part number and revision are listed
- Note any firmware or driver requirements
Common HCL failures: RAM that works in the server but causes intermittent errors under load (speed/timing mismatch), storage controllers that aren't recognized after firmware updates, and expansion cards that conflict with the server's PCIe topology.
⚠️ Exam Trap: HCL compliance is a procurement-time check, not a post-installation fix. If you discover a component isn't on the HCL after installation and the server is behaving strangely, the correct answer is "replace with HCL-approved component," not "update drivers."
Reflection Question: You install a third-party RAM module that appears to work initially but causes random crashes under peak load. What is the most likely cause, and what is the correct resolution?
