FCFS and SSTF Disk Scheduling

Why the Order of Disk Requests Actually Matters

A mechanical hard disk's read/write head must physically move across the disk's surface to reach different data - unlike RAM, where any address is equally fast to reach, a disk's access time depends heavily on how far the head must physically travel. Disk scheduling decides the order in which pending requests are serviced to minimize this movement.

FCFS Disk Scheduling: Simply Service Requests in Arrival Order

Head starts at track 50
Request queue (in arrival order): 95, 20, 80, 10

Head movement: 50 -> 95 -> 20 -> 80 -> 10
Total head movement: 45 + 75 + 60 + 70 = 250 tracks

Just like FCFS process scheduling (covered earlier in this section), FCFS disk scheduling simply services requests in the exact order they arrive - simple and fair in that sense, but as shown above, it can cause a huge amount of unnecessary head movement by jumping back and forth erratically.

SSTF: Always Service the Closest Request Next

Head starts at track 50
Request queue: 95, 20, 80, 10

Head movement: 50 -> 80 (closest) -> 95 (closest remaining) -> 20 (closest remaining) -> 10
Total head movement: 30 + 15 + 75 + 10 = 130 tracks

Shortest Seek Time First (SSTF) always services whichever pending request is physically closest to the head's current position - dramatically reducing total head movement compared to FCFS in this example.

SSTF's Real Weakness: Potential Starvation

If requests keep arriving near the head's current position, a request sitting far away can be repeatedly passed over and effectively starved - the exact same starvation problem covered in the CPU scheduling section, just showing up here in a disk-scheduling context instead.

Why Neither of These Is the Final Answer

SSTF solves FCFS's inefficiency but introduces a real fairness problem - motivating the SCAN and LOOK family of algorithms, covered in the next two notes, which aim to balance both efficiency and fairness together.