ScholarGate
Assistant

Secondary Storage Devices

Secondary storage devices — magnetic hard disks and flash-based solid-state drives — hold data persistently outside main memory, with very different performance characteristics that shape how systems store and access information.

Definition

Secondary storage devices are non-volatile storage media — principally magnetic hard disk drives and flash solid-state drives — used to retain data persistently across power cycles, providing large capacity at lower cost and lower speed than main memory.

Scope

This topic covers the technologies and performance of persistent storage: magnetic disk geometry, seek and rotational latency, and disk scheduling; NAND flash and solid-state drives, including their access behavior, wear, and the flash translation layer; and how these characteristics affect system design. It treats the devices themselves. It excludes redundancy and reliability arrays (RAID and storage reliability) and the file-system software above them (file systems).

Core questions

  • What determines the latency and throughput of a magnetic hard disk?
  • How does flash-based solid-state storage differ from magnetic disk in behavior and performance?
  • Why do solid-state drives need a flash translation layer and wear leveling?
  • How do storage device characteristics influence system and software design?

Key concepts

  • magnetic disk geometry (tracks, sectors, cylinders)
  • seek time and rotational latency
  • disk scheduling
  • NAND flash and solid-state drives
  • flash translation layer
  • wear leveling
  • throughput and latency
  • persistence and non-volatility

Mechanisms

A magnetic disk stores data on spinning platters; access requires moving the head to the right track (seek) and waiting for the sector to rotate underneath (rotational latency), so disk scheduling reorders requests to reduce head movement. A solid-state drive stores data in NAND flash, which is read and written in pages but erased in larger blocks and wears out with use; a flash translation layer maps logical addresses to physical pages and performs wear leveling and garbage collection.

Clinical relevance

Storage device characteristics often dominate the performance of data-intensive systems. The transition from mechanical disks to solid-state drives dramatically reduced access latency and reshaped database, file-system, and operating-system design, while the distinct behavior of flash — erase-before-write, wear, and garbage collection — continues to influence how storage software is built.

History

The magnetic hard disk, introduced by IBM in 1956, dominated secondary storage for half a century, improving steadily in density and cost. NAND flash memory, commercialized from the late 1980s, enabled solid-state drives that displaced disks in many roles from the 2000s onward, trading higher cost per bit for far lower latency and no moving parts.

Key figures

  • John L. Hennessy
  • David A. Patterson
  • Bruce Jacob

Related topics

Seminal works

  • hennessy2019
  • jacob2008

Frequently asked questions

Why are solid-state drives faster than hard disks?
Hard disks must mechanically move a head and wait for platters to rotate, incurring milliseconds of latency. Solid-state drives access flash electronically with no moving parts, so their latency is far lower and they handle random access much better, though they are more expensive per byte.
What is wear leveling?
Flash memory cells can endure only a limited number of erase-write cycles. Wear leveling spreads writes evenly across all cells, via the flash translation layer, so no region wears out prematurely, extending the usable life of a solid-state drive.

Methods for this concept

Related concepts