What is SSD (solid-state drive)?
A storage device that stores data using NAND flash memory and a controller. It has no moving parts and offers faster random access and better shock resistance than an HDD.
Japanese name: SSD
SSD (solid-state drive) appears in 11 practice questions across 3 topics, including SRAM and flash memory, I/O Interfaces & Buses, Performance Analysis.
Field: SRAM and flash memory
What is inside an SSD
An SSD (solid-state drive) is a storage device with several NAND flash memory chips, a controller that manages them and, in many cases, DRAM that holds the address mapping table, all on one board. The controller handles the FTL, which maps logical addresses to physical locations in flash, ECC to correct errors, wear leveling to spread writes across the whole device, and garbage collection to erase blocks that are no longer needed and create free space. By running many chips in parallel over multiple channels, it achieves higher transfer rates than a single chip could.
Interfaces and form factors
SSDs connect to the host through SATA, the same interface as HDDs, or NVMe, which runs over PCI Express. SATA is limited to 6 Gb/s by specification, and actual transfer rates stay in the 500 MB/s range. NVMe SSDs use several PCI Express lanes (usually x4) and, depending on the generation, can reach transfer rates of several GB/s; they can also process many commands in parallel, which keeps latency low. Form factors include the same size as a 2.5-inch HDD, the narrow M.2 board used in laptops and elsewhere, and form factors for data centers.
SSD vs. HDD
An HDD reads and writes by moving a head over spinning magnetic platters, so accessing a distant location takes several milliseconds. An SSD has no moving parts and responds to random reads in roughly tens to 100 µs, making it especially fast for workloads with many small files. It also uses less power, withstands shock and vibration and makes no noise. On the other hand, large-capacity HDDs are generally cheaper per unit of capacity, and data centers also use HDDs to store large volumes of data. An HDD can overwrite in place, but flash must be erased before it is written, which complicates an SSD's internal processing.
Write endurance and data retention
Each rewrite degrades the insulating films in NAND flash cells, so SSDs have a guideline for total writable data, expressed as TBW (terabytes written) or DWPD (drive writes per day). TLC and QLC, which store more bits per cell, generally have lower rewrite limits. TRIM, which lets the host report deleted data, and spare area hidden from the user (over-provisioning) reduce write amplification and extend lifetime. How long data survives with power off varies greatly with wear and storage temperature, and a heavily worn SSD stored hot for a long time can lose data.
FAQ
Which is faster, an SSD or an HDD?
An SSD is generally faster, especially for random access: an HDD needs several milliseconds to move its head, while an SSD responds in roughly tens to 100 µs. Large HDDs are cheaper per unit of capacity, so the two are used for different purposes.
What is the difference between NVMe and SATA SSDs?
SATA is an interface designed for HDDs and is limited to 6 Gb/s by specification. NVMe is a protocol for flash that runs over PCI Express, and multiple lanes plus parallel processing of many commands give higher transfer rates and lower latency.
Does an SSD keep its data while powered off?
NAND flash memory is nonvolatile, so data remains when power is off. How long it lasts depends on wear from rewriting and on storage temperature, however, and a worn SSD stored hot for a long time can lose data.