How Solid State Drives Work
What Is an SSD?
Solid state drives, or SSDs, function by using transistors and other solid-state electronics to store and retrieve data. Most employ special floating-gate field effect transistors, or FETs, which change their conductivity in response to external voltages. Unlike the majority of FETs, however, these devices feature electrically isolated, or floating, gate sections. Any charge that builds up on the gate has nowhere to go and thus retains its state for some time.
Individual FETs, or cells, act like NAND logic gates, with each storing a charge or lack thereof to represent a single 0 or 1 and the device only outputting a 0 if all of its inputs are 1s. These transistors include source lines and bit lines that connect each cell to control reading and writing and arrange the individual bits into grids. Additional transistors are employed to control addressing, or which bit is being accessed at a given moment.
Each grid, or block, is further divided into rows of bits known as words, whose sizes typically max out at around 16K. With typical block sizes of up to 256 pages, common page sizes can reach around 4MB.
While RAM generally beats all other devices in terms of low read and write access delays, NAND memory consistently outpaces hard drives in these categories, and the fact that seeking is unnecessary is a clear advantage. The one operation that flash falls behind on is erasure, which isn’t needed for hard drives or RAM.
Erasing stored bits at the block level is the best way to minimize the negative side effects of the high voltages required for erasure, so these operations take some time. As SSDs become increasingly full, they may also slow down, as they have to locate blocks that are OK to delete and erase them completely before writing new pages.
Techniques like garbage collection, where blocks are modified by writing their new state to empty blocks before deleting the originals, and TRIM, or skipping rewrite operations for select data when block erasure is performed, can decrease overhead. Other methods, like wear leveling, ensure that different NAND blocks are generally used the same amount to increase overall drive life. Wear leveling, garbage collection and TRIM also help mitigate the impacts of having to write whole blocks just to erase select pages.
What Are the Parts in an SSD?
SSDs revolve around two main components: the memory and the controller. The memory consists of the blocks of NAND flash gates that actually hold the data. The controllers are complex electronic circuits that perform read, write and erasure operations.
Controllers also take care of monitoring blocks for wear leveling, finding unusable sectors that have sustained damage, data encryption, error correction and garbage collection. Different manufacturers employ a range of proprietary techniques to give their SSD products competitive edges, and these typically revolve around how controllers function.
How Does a Hard Drive Work?
Hard drives employ magnetic storage to accomplish read and write operations. Each disk inside a drive contains distinct sections of paramagnetic materials, or substances that change their magnetic field in response to external magnetic fields. When a read-write arm uses an electromagnet, or head, to set the orientation of a given field, that bit becomes a 0 or a 1 that can later be retrieved.
One significant difference between hard drives and solid-state technology is that hard drives require moving parts. Voice coils, motors and other hardware work together to keep the drives spinning and shift the arm back and forth so that its read-write head can access individual data sectors. This traditional methodology has some advantages, like established market share and well-known failure modes, but compared to SSDs, it generally suffers from heightened wear, slower access and increased environmental susceptibility.