§37.1–37.3The Interface … A Simple Disk Drive

Now the star device of persistence: the hard disk drive. File systems are shaped by how disks behave, so we build a model of one first.

37.1 The Interface

A disk is simply an array of sectors sector The atomic unit of a hard disk: a 512-byte block. Sectors are numbered 0…n−1, forming the disk's address space. Manufacturers guarantee only that a single-sector write is atomic; a larger write interrupted by power loss can leave a partial 'torn write.' defined in ch. 37 — open in glossary — 512-byte blocks numbered 0 to n−1 (the drive’s address space). You can read or write any sector; file systems often move 4KB (8 sectors) at a time. But the only atomicity guarantee is that a single-sector write completes fully or not at all — a larger write cut short by power loss can leave a partial torn write. And the “unwritten contract”: nearby blocks are faster to reach than distant ones, and sequential access is far faster than random.

37.2 Basic Geometry

Data lives on a platter platter A circular hard surface (aluminum coated with a thin magnetic layer) that stores bits persistently, even when the drive is off. A disk has one or more platters, each with two surfaces, all bound to a spindle that a motor spins at a fixed rate (RPM). defined in ch. 37 — open in glossary — a circular surface coated in magnetic material, holding bits even when powered off. Platters bind to a spindle spun by a motor at a fixed rate (7,200–15,000 RPM). Each surface holds thousands of concentric tracks track A concentric circle of sectors on a platter surface. A single surface holds many thousands of tightly packed tracks (hundreds fit within a human hair's width). A read/write head, mounted on a movable disk arm (one head per surface), positions over the desired track. defined in ch. 37 — open in glossary (hundreds fit in a hair’s width); a read/write head on a movable arm (one per surface) positions over the target track.

37.3 A Simple Disk Drive

Servicing a request takes three steps — walk through a real one:

Two costs dominate. The rotational delay rotational delay The time a disk spends waiting for the desired sector to rotate under the head. On average it is about half of a full rotation (R/2). Together with seek time, one of the two dominant costs of a disk I/O. defined in ch. 37 — open in glossary is the wait for the sector to spin under the head (avg ≈ R/2). The seek time seek time The time to move the disk arm so the head is over the target track. It has phases — acceleration, coasting, deceleration, and settling (the settling alone is 0.5–2 ms). Along with rotation, one of the most costly disk operations; the AVERAGE seek is about one-third of a full-stroke seek. defined in ch. 37 — open in glossary is moving the arm to the right track — and it’s not instant: the arm must accelerate, coast, decelerate, and carefully settle (0.5–2 ms) onto the track. Then the transfer completes the I/O.

A few more realities

Real drives add cleverness. Track skew track skew An angular offset in sector numbering between adjacent tracks, arranged so that after switching tracks a sequential read finds the next block just arriving under the head — rather than having just missed it and paying nearly a full rotation. defined in ch. 37 — open in glossary shifts the sector numbering between adjacent tracks so that after a track switch the next sequential block is just arriving — not just missed (which would cost a whole rotation):

Two more: a multi-zoned disk multi-zoned disk A disk organized into zones (consecutive sets of tracks) where outer zones pack MORE sectors per track than inner zones — because there is simply more room on the outer, longer tracks. Each zone has a fixed number of sectors per track. defined in ch. 37 — open in glossary packs more sectors on outer tracks (there’s more room out there), grouping tracks into zones with a fixed sectors-per-track. And every drive has a track buffer track buffer A small memory inside the drive (a disk cache, historically the track buffer, usually 8–16 MB) that holds data read from or written to the platters — e.g. on reading one sector the drive may cache the whole track to answer later same-track requests quickly. defined in ch. 37 — open in glossary — a small (8–16 MB) internal cache; e.g. on a read it may pull in the whole track to answer later same-track requests fast. That cache forces a write policy choice — write-back caching write-back caching A disk-cache write policy (a.k.a. immediate reporting) that acknowledges a write as soon as the data reaches the drive's memory — making the drive appear faster, but risking correctness if a crash occurs before the data hits the platter (dangerous when order matters, e.g. journaling). The safe alternative is write-through, which acks only after the data is actually on disk. defined in ch. 37 — open in glossary vs write-through:

With this model of where time goes on a disk, we can now put numbers to it.