Disk Scheduling in Modern Systems
You'll learn how solid-state drives (SSDs) have fundamentally changed disk scheduling and why traditional algorithms like SCAN and SSTF matter far less today.
The Old Rules No Longer Apply
Traditional disk scheduling algorithms—FCFS, SSTF, SCAN, and their variants—were designed around one fundamental constraint: the mechanical seek time of spinning hard disk drives (HDDs). Moving a physical read/write head across platters was expensive, so minimizing seek distance was crucial.
SSDs changed everything. They have no moving parts. Reading from "block 1" versus "block 10,000" takes essentially the same time because there's no physical head to reposition. This is like the difference between walking across a library to grab a book (HDD) versus instantly teleporting to any shelf (SSD).
What Modern Storage Cares About
With SSDs, scheduling priorities shift dramatically:
- Seek time optimization becomes irrelevant — the core problem traditional algorithms solved simply doesn't exist
- Wear leveling matters more — spreading writes evenly across memory cells to prevent premature failure
- Write amplification becomes a concern — internal operations that multiply the actual data written
- Parallelism is key — modern SSDs have multiple channels and can handle concurrent operations efficiently
Many operating systems now use simpler schedulers like "noop" (no operation) or deadline-based schedulers that focus on fairness and latency rather than minimizing physical movement.
The Hybrid Reality
Some systems still use HDDs for bulk storage, so algorithms like SCAN haven't disappeared entirely. But for performance-critical workloads on SSDs, the OS often just submits I/O requests in roughly the order received, letting the SSD's internal controller handle optimization.
Key Takeaway: SSDs eliminated the mechanical seek-time bottleneck that traditional disk scheduling algorithms were designed to solve, making geometric optimization largely obsolete in favor of fairness, wear leveling, and parallelism.