RAID, Storage & Motherboard Master Notes

RAID Fundamentals

• Definition: Redundant Array of Independent (‘Inexpensive’) Disks – grouping multiple physical drives so the operating system sees one logical volume.
• Why RAID?
  • Increase speed (striping)
  • Increase fault-tolerance (mirroring/parity)
  • Increase capacity (combining drives)
• Redundancy ≠ Backup
  • Redundancy keeps the service running after a component failure (e.g. losing a disk or power supply).
  • Backup is an offline/off-site copy used after a total loss (fire, flood, ransomware). Example: tape libraries swapped monthly.
  • Analogy: carrying two packs of Skittles every day (redundancy) vs. being able to drive to the store and buy a new pack (backup).

Real-World Hardware Context

• Enterprise storage racks may contain $20+$ drives (each $2\text{–}10\,\text{TB}$). Classroom diagrams are simplified to 4-disk examples.
• NAS on a desk ≈ 4–8 bays, same RAID concepts, smaller scale.
• Hot-swap bays: red tab → pull drawer → slide drive out while system stays online.

RAID Levels Discussed In Class

• RAID 0 (Striping)

  • Purpose: maximum speed & full capacity.
  • Method: data blocks split across all disks ("RAID | is | fun" → disk 1 gets RAID, disk 2 gets is fun).
  • Fault Tolerance: $0$. Lose one drive → array dead → data lost.
  • Capacity: sum of all drives (e.g. 2 × $5\,\text{TB} = 10\,\text{TB}$ usable).
  • Mnemonic: "RAID 0 = 0 redundancy".

• RAID 1 (Mirroring)

  • Purpose: simple, highly reliable storage; speed not main goal.
  • Method: every block duplicated on a second drive (full mirror).
  • Fault Tolerance: can lose up to half the disks (one mirror from each pair) with no downtime.
  • Capacity: $\tfrac12$ of raw (buy two drives for every one needed).
  • Cost: $100\%$ redundancy overhead; perfect for critical but low-activity systems (e.g. bank transaction logs).

• RAID 5 (Striping + Distributed Parity)

  • Purpose: balance of speed, capacity & redundancy; common in SMB servers.
  • Method: data striped like RAID 0, plus 1 full disk equivalent of parity distributed across all drives.
  • Parity: mathematical snapshot ("RIF" in example) that can rebuild one missing block using remaining data $+$ parity.
  • Fault Tolerance: lose any one drive and keep running.
  • Capacity: $\text{(n – 1)}$ drives. 4 × $5\,\text{TB}$ ⇒ $15\,\text{TB}$ usable.
  • Cost Efficiency: only one extra disk regardless of array size; cheapest way to add redundancy.
  • Performance: slower than RAID 0 because parity must be written each time.

• RAID 6 (Striping + Dual Parity)

  • Purpose: like RAID 5 but safer; survives two simultaneous disk failures.
  • Capacity: $\text{(n – 2)}$; parity cost = 2 drives.
  • Performance: slightly slower than RAID 5 (extra parity math).

• ## RAID 10 (1 + 0 = Mirrors of Stripes)

  • Purpose: highest speed and highest fault-tolerance (if budget allows).
  • Method: first stripe data, then mirror each stripe set.
  • Fault Tolerance: can lose one disk in every mirrored pair (up to half the array) with no downtime.
  • Capacity & Cost: same $50\%$ overhead as RAID 1 (need double the disks) plus a dedicated RAID controller.
  • When chosen: “Money no object, no unscheduled downtime” scenarios—banking, high-traffic DB, virtualization clusters.

Tape Backup & Disaster Recovery

• Tape drives hold $\ge 10\,\text{TB}$ per cartridge; very slow but extremely cheap per TB.
• Typical practice: nightly incremental copy → monthly cartridge rotation → off-site vault (tornado scenario).
• After catastrophic loss: rebuild RAID from replacement disks, then restore from tape.

Key Exam Numbers & Formulas

• RAID 0 capacity: $\sum \text{drive sizes}$
• RAID 1/10 capacity: $\tfrac12\times\sum$ (mirrored)
• RAID 5 capacity: $(n-1)$ drives
• RAID 6 capacity: $(n-2)$ drives
• Example question: Need $20\,\text{TB}$ usable, RAID 5, drives $5\,\text{TB}$ each → $\frac{20}{5}+1 = 5$ disks total (4 for data, 1 parity).

Motherboard Form Factors

• ATX – $12\,\text{in} \times 9.6\,\text{in}$ rectangle (mainstream desktops/gaming rigs).
• Micro-ATX – $9.6\times 9.6$ square (budget / small office PCs).
• Mini-ITX – $6.7\times 6.7$ square (HTPC, thin clients, firewalls).
  • Remember: identical numbers ⇒ square; differing numbers ⇒ full ATX.

Expansion Slots

• PCI vs. PCIe (Express)
  • PCIe is faster; uses serial “lanes”. Sizes: $\times1, \times4, \times8, \times16$ (71 pins at full length).
  • Smaller card fits in larger slot (works at its own lane count).
  • Graphics cards usually require $\times16$ and extra 6-/8-pin power.
• AGP appears only on legacy hardware (pre-2007).

Power Connectors

• 24-pin (20 + 4) ATX main power → motherboard.
• 4-/8-pin EPS12V → CPU socket area.
• 6-/8-pin PCIe → high-end GPUs.
• SATA power (15-pin) vs. SATA data (7-pin) for drives.
• Front-panel header (power switch, reset, HDD LED); mis-wiring = PC won’t start.

Storage Interfaces

• SATA III – $6\,\text{Gb/s}$ cables; HDD & 2.5" SSD.
• M.2 slot – accepts stick-style SSDs.
  • If drive uses NVMe protocol: ≈10× faster than SATA, communicates over PCIe lanes.
• eSATA – external SATA enclosure.

CPU Sockets & Installation

• LGA (Land Grid Array) – pins on motherboard (typ. Intel).
• PGA (Pin Grid Array) – pins on CPU (typ. AMD).
• ZIF lever secures chip; align golden triangle → apply pea-sized thermal paste → attach cooler.

Cooling & Airflow

• Active heatsink = fins + fan.
• Passive heatsink = fins only (ultrabooks, ARM devices).
• Liquid AIO: water block on CPU, radiator/fans mounted to case; custom loops for GPU/CPU.
• Correct flow: front/bottom = intake (cool air); top/rear = exhaust (hot air rises).
• Symptom: PC runs few minutes then shuts off → check fan power, thermal paste, dust (compressed-air maintenance).

BIOS vs. UEFI

• BIOS: legacy 16-bit interface, keyboard only, uses MBR partition scheme (max 4 primary partitions, \le $2\,\text{TB}$ each).
• UEFI: graphical/mouse, secure boot, supports GPT (≈$128$ partitions, $\gt2\,\text{TB}$ drives).
• Boot order: change to USB/DVD for OS install, then revert; wrong order ⇒ "Operating system not found".
• Secure Boot: verifies digitally signed loaders; disable if installing unsigned Linux.
• BIOS passwords
  • User/Boot – stops OS start.
  • Supervisor – locks BIOS settings.

Security & Virtualization Features

• TPM 2.0 chip (Windows 11 requirement)
  • Stores BitLocker keys, DRM secrets; failure to save recovery key = data lost.
• HSM – external enterprise version of TPM for servers/banks.
• Intel VT-x / AMD-V – virtualization extensions; enable in BIOS for Hyper-V, VMware, etc.

Fan & Environmental Monitoring

• BIOS/UEFI can show RPM & temperature; RGB lighting can color-code temps.
• Keep blanking plates in unused PCI slots for proper airflow.

PC-Build Decision Scenarios (Exam Style)

• Gaming rig: dedicated GPU, $\ge16\,\text{GB}$ RAM, $\ge1000\,\text{W}$ PSU, liquid cooling, surround audio.
• Office/VM host: onboard video, $24\,\text{GB}+$ RAM, $750\,\text{W}$ PSU, KVM or dual-monitor output, external backup drive.

Exam Tips & Gotchas

• "Recently upgraded CPU and PC shuts down after 5 min" → forgot to connect CPU-fan or applied no thermal compound.
• SATA ports often color-coded differently; still key on the small "L" shape for identification.
• Loopback test (ping $127.0.0.1$) verifies NIC driver before cabling.
• AGP, ISA, 8086 references signal legacy distractors.
• Always read RAID capacity/fault tolerance as usable vs. raw.