Technical guide

How Storage Works

What is inside a drive, the link it sits on, and the ways stored data is placed, protected and recovered.

A series on what is inside drives and the links and protections around them, in eight parts. Part 1 lays out the memory and storage hierarchy, from registers to tape, and the rule that decides where data lives. Part 2 opens a hard drive: heads and platters, conventional and shingled recording, helium, energy-assisted recording and drives with two actuators. Part 3 covers tape and cold archive, with LTO tape and the optical, glass and DNA media still finding their place. Part 4 covers PCIe Gen6 and Gen7 for storage: the flit, error correction, L0p, and the power and heat a drive is held to. The parts after it cover how a host tells a drive where data goes, and how RAID, encryption and recovery protect what is stored. How the flash inside an SSD works is in Flash Foundations, and choosing and setting up drives is in Storage and Memory in Practice.

The series

Read in order

Each part stands on its own.

The memory and storage hierarchy: bandwidth from a hard drive to one GPU’s HBM, the five-minute rule from 1987 to 2017, and one SSD’s write speed through its phasesA three-panel schematic. Left: two bars on a log scale, a hard drive at 270MB/s and one GPU’s HBM at 19.2TB/s, the HBM bar far longer. Center: a line of the DRAM to hard drive break-even interval from 400 seconds in 1987 to 4 hours in 2017 on a log scale, with a dashed line at five minutes. Right: a line of one SSD’s write speed against data written, high while the cache absorbs writes, then a slower phase, then a steady middle speed.GUIDE / HOW STORAGE WORKS, PART 1Every memory wins on one axisBANDWIDTH · FIVE-MINUTE RULE · WRITE SPEED01 BANDWIDTHHDD 270MB/sHBM 19.2TB/slog scale02 BREAK-EVEN5 min400s4h1987201703 WRITE SPEEDcache, fold, then TLC

Part 1

The Memory and Storage Hierarchy

Nine memories from registers to tape on latency, bandwidth, capacity and cost, and the rule that decides where data lives.

Read Part 1
How a hard drive works: conventional and shingled tracks, a laser heating the media under the write head, and random reads per terabyte on one actuator and on twoA three-panel schematic. Left: two strips of tracks, conventional tracks with gaps between them above and shingled tracks that overlap below. Center: a write head over a row of magnetic grains with a laser heating one spot to near 450 degrees Celsius. Right: two bars of random reads per terabyte, 8.4 for one actuator and 16.9 for two.GUIDE / HOW STORAGE WORKS, PART 2More bits, more reads, one driveTRACKS · HEAT · ACTUATORS01 TRACKSCMR, side by sideSMR, overlapped02 HEATnear 450°Claser-heated spot03 ACTUATORS8.416.9onetworeads per TB

Part 2

How a Hard Drive Works

Heads, tracks, CMR and SMR, helium, HAMR and ePMR, and dual actuators: how a hard drive stores and finds data, and what its spec sheet labels mean.

Read Part 2
Tape and cold archive: a tape drive writing 32 tracks per pass on a 40TB cartridge, a 5.5TB optical cartridge beside a 4.84TB glass platter, and DNA’s write rate of 0.001MB/s against tape’s 400MB/sA three-panel schematic. Left: a zigzag path of one tape head passing back and forth, labeled 32 tracks per pass and a 40TB native cartridge. Center: a cartridge of eleven upright discs labeled 5.5TB above a glass platter drawn as stacked layers labeled 4.84TB. Right: two bars on a log scale, DNA at 0.001MB/s a short stub and tape at 400MB/s nearly the full width.GUIDE / HOW STORAGE WORKS, PART 3Four ways to keep data coldTAPE · OPTICAL AND GLASS · DNA01 TAPE32 tracks per pass40TB native cartridge02 OPTICAL, GLASS5.5TB disc cartridge4.84TB glass platter03 DNATape 400MB/sDNA 0.001MB/swrite rate, log scale

Part 3

Tape and Cold Archive

How LTO tape stores data and why it still wins on cost and power at rest, then what optical archive, glass and DNA storage are today and what is still research.

Read Part 3
PCIe Gen6 for storage: a flit with payload and link fields, a link narrowing to fewer active lanes, and a temperature scale with throttle marksA three-panel schematic. Left: a wide teal payload field followed by three thin copper link fields, above a row of small flit slots. Center: a drive box and a host box joined by four lanes, two solid teal and two dashed and idle. Right: a vertical temperature scale with three marks and a copper band between the upper two.GUIDE / HOW STORAGE WORKS, PART 4PCIe Gen6 for storageFLIT · LANES · HEAT01 FLIT02 LANES03 HEAT

Part 4

PCIe Gen6 and Gen7 for Storage

What PCIe Gen6 changes for SSDs: the 256-byte flit and its cost, FEC and replay, L0p, OCP power and heat limits, the first Gen6 drives, and PCIe 7.0 and 8.0.

Read Part 4

All 8 parts

  1. Part 1The Memory and Storage HierarchyPublished
  2. Part 2How a Hard Drive WorksPublished
  3. Part 3Tape and Cold ArchivePublished
  4. Part 4PCIe Gen6 and Gen7 for StoragePublished
  5. Part 5ZNS and FDP: Telling the Drive Where Data GoesIn progress
  6. Part 6RAID, Erasure Coding and Checksumming FilesystemsIn progress
  7. Part 7Self-Encrypting DrivesIn progress
  8. Part 8Data Recovery and RetentionIn progress

On this site

Where it connects

The guides link these pages where they lean on them.