2.2.2. RAID Levels and Types
💡 First Principle: RAID (Redundant Array of Independent Disks) solves the reliability problem of individual drives by combining multiple drives into a single logical unit. Each RAID level makes different choices about how to distribute data and parity across drives—which determines its performance, capacity efficiency, and fault tolerance.
Think of RAID like the backup strategy for your storage medium itself. Just as you choose a backup method based on RTO/RPO requirements, you choose a RAID level based on performance, capacity, and fault tolerance requirements.
RAID Levels — The Full Breakdown
| RAID Level | Min Drives | Fault Tolerance | Usable Capacity | Performance | Use Case |
|---|---|---|---|---|---|
| RAID 0 | 2 | None | 100% | Highest read/write | Temp data, scratch space — NOT for production data |
| RAID 1 | 2 | 1 drive | 50% | Better read (both disks serve reads); normal write | OS drives, critical small datasets |
| RAID 5 | 3 | 1 drive | (N-1)/N | Good read; slower write (parity calc overhead) | General-purpose workloads |
| RAID 6 | 4 | 2 drives | (N-2)/N | Good read; slowest write (double parity) | Large arrays, high-value data |
| RAID 10 | 4 | 1 per mirror pair | 50% | Excellent read/write | High-performance databases |
| JBOD | 1+ | None | 100% | Pass-through | Simple span, no redundancy |
Capacity Calculation Examples (for exam scenarios):
- 4 × 2TB drives in RAID 5: (4-1) × 2TB = 6TB usable
- 4 × 2TB drives in RAID 6: (4-2) × 2TB = 4TB usable
- 4 × 2TB drives in RAID 10: 50% of 8TB = 4TB usable
- 4 × 2TB drives in RAID 0: 4 × 2TB = 8TB usable (no redundancy)
Hardware vs. Software RAID
- Hardware RAID: Dedicated RAID controller card with its own CPU and cache. Higher performance, OS-independent (the OS sees a single logical drive). Cache is typically battery-backed to protect write data during power loss.
- Software RAID: OS manages parity and striping using the system CPU. Lower cost, more flexible, but consumes system CPU and RAM resources. Examples: Linux MD RAID, Windows Storage Spaces.
Capacity Planning with RAID
When planning storage, you must account for: usable capacity after RAID overhead, a hot spare (a drive kept ready to automatically begin rebuilding when a drive fails), and growth headroom. A 60–70% utilization ceiling is a reasonable operational rule—drives near capacity suffer performance degradation and leave no room for unexpected growth.
⚠️ Exam Trap: RAID is not a backup. RAID protects against drive failure, but it does not protect against file deletion, ransomware, data corruption, or site-level disasters. The exam specifically tests this distinction—RAID and backup are complementary, not equivalent.
Reflection Question: A company needs storage for a database that requires high write performance and can tolerate one drive failure, using exactly 4 drives. Which RAID level is appropriate, and what is the usable capacity if each drive is 4TB?
