2.2.2. Storage Devices and RAID Configurations

💡 First Principle: The choice of storage technology (e.g., NVMe vs. SATA vs. HDD) and configuration (e.g., RAID level) is a critical trade-off between speed, capacity, cost, and data redundancy.

Scenario: A video editor complains their workstation is "stuttering" when scrubbing 4K video timelines. Their OS and applications are on a fast NVMe SSD, but their large video project files are on a single large HDD. You theorize the HDD is the bottleneck.

Storage is where the operating system, applications, and user data reside. The type and configuration of storage are critical for both performance and data safety.

• Storage Devices:
  • Hard Disk Drives (HDD): Mechanical drives with spinning platters. Offer high capacity at a low cost per GB but are much slower than SSDs, and their moving parts make them more prone to mechanical failure. Files stored in non-contiguous pieces (fragmentation) slow a mechanical drive because the head must seek between the pieces; defragmentation helps HDDs but not SSDs. Common form factors are 3.5-inch (desktops) and 2.5-inch (laptops).
  • Solid-State Drives (SSD): Use flash memory with no moving parts (silent and shock resistant, but a higher cost per GB) for much faster performance. Common interfaces include SATA (looks like a 2.5-inch HDD) and NVMe (Non-Volatile Memory Express), which uses the PCIe bus via an M.2 slot for the highest speeds. The M.2 slot is on the motherboard itself (no drive bay or cable), which suits compact builds; NVMe vs SATA is the protocol the drive uses.
  • Hybrid Drives (SSHD): Combine a spinning-platter HDD with a small flash (NAND) cache to speed up frequently used data.
  • Drive Interfaces and Form Factors: Internal drives connect through SATA, SAS (Serial Attached SCSI, the enterprise/server interface, with its own connector and controller, built for performance and reliability) or NVMe/M.2; external drives attach through USB, Thunderbolt or eSATA. SATA III is a 6 Gbps physical interface for 2.5-inch/3.5-inch drives (and some M.2 SSDs). NVMe is a logical protocol that runs over PCIe lanes. M.2 is a physical form factor (a small stick) that can carry either SATA or NVMe SSDs, depending on its keying. eSATA is a legacy external SATA connector for outboard drives, largely replaced by USB 3.x and Thunderbolt.
  • Replacing a SATA Drive: Power down, unplug the power cord and discharge static (ESD strap); disconnect the SATA data and SATA power cables; unscrew and remove the failed drive, then install and secure the replacement; reconnect data and power, restore power, and initialize/format the new drive in the OS.
  • Removable Storage: USB flash drives, memory cards.

Visual: M.2 SSD (right) vs mSATA SSD (left) — M.2 NVMe offers the highest speeds via the PCIe bus, while SATA SSDs use the same interface as traditional hard drives.

• Drive Configurations (RAID - Redundant Array of Independent Disks): RAID combines multiple physical drives into a single logical unit to improve performance, add fault tolerance, or both. The four levels you must know:
  • RAID 0 (Striping): Splits data across two or more drives for maximum read/write speed and full combined capacity. Offers no redundancy — if any single drive fails, the entire array is lost. Minimum 2 drives.
  • RAID 1 (Mirroring): Writes an identical copy of all data to two drives. Survives the failure of one drive, but usable capacity is only half the total. Minimum 2 drives.
  • RAID 5 (Striping with Parity): Stripes data across three or more drives and distributes parity information so the array can rebuild from a single drive failure. Balances speed, capacity, and redundancy; usable capacity is total minus one drive. Minimum 3 drives. After one drive fails the array stays online in degraded mode: data is rebuilt from parity on the fly (slower) and nothing is protected until the drive is replaced and the array rebuilds; a second failure loses the array.
  • RAID 10 (a "stripe of mirrors"): Combines mirroring and striping by striping data across mirrored pairs. Delivers both high performance and fault tolerance, and can survive multiple drive failures as long as no full mirrored pair is lost. Minimum 4 drives; usable capacity is half the total.

⚠️ Common Pitfall: Confusing RAID with a backup. RAID provides redundancy against hardware failure, but it does not protect against accidental deletion, malware, or catastrophic events like fire or theft. A separate backup is still essential.

Key Trade-Offs:

• Performance (RAID 0) vs. Redundancy (RAID 1/5/10): You must choose a RAID level that balances the need for speed with the importance of protecting the data against drive failure.

Reflection Question: For the video editor scenario, what two-part storage solution would you propose to fix the performance bottleneck for active projects while also providing a redundant solution for archiving completed work?