Traditional hard disk drives work via the principles of magnetic data storage. A device called an actuator arm with a read-and-write head attached moves across the surface of spinning magnetic disks called platters to retrieve data stored in different locations. The heads, just nanometres from the surface, read and change the pattern of magnetised particles beneath them. These particles have different magnetic polarities - positive or negative, and these represent either a 0 or a 1 in binary code.
This design means that there is always a delay or lag before information can be retrieved. The arm must move to the correct location on the disk for a specific piece of data, and it may have to repeat this process several times before a particular file or application can be fully assembled in RAM and opened. This lag - called latency - is normally measured in milliseconds but it can climb to whole seconds if the disk is spinning up from a low power rest mode. By contrast, the speed of modern CPUs is measured in nanoseconds. One nanosecond is equal to one million milliseconds.
As each generation of CPU became faster and faster, the gap between hard drive latency and CPU speed grew greater and greater, despite improvements in hard drive technology enabling the platters to spin more quickly. The inherent limitations of HDD technology meant platter revolution eventually maxed out at around 100,000 revolutions per minute (RPM).
Solid-state drives work differently. They rely on much faster flash memory which does not require any moving parts to operate. Data is stored within cells that operate as floating (electronically isolated) logic gates. Logic gates are computational switches that produce a single output from two binary inputs. Electrons interact with these gates, changing their charge from one state to another in a flash. This change in charge rapidly erases the existing data in the cell so new data can be entered. A second gate, called a control gate, regulates the entry and exit of data.
These electrical charges are retained even when the device is not connected to power. For this reason, flash is considered non-volatile memory.
The memory cells themselves are a type of metal–oxide–semiconductor field-effect transistor or MOSFET.
There are two types of flash memory - NOR or NAND, so-called for the type of logic gate they employ. NAND flash is the most widely used and has become the default choice in many familiar devices including USB drives and modern computers. It is less expensive and more capacious than NOR flash.
NAND memory is laid out in grids, called blocks, each made of rows called pages. A typical NAND block varies between 256 kilobytes and 4 megabytes in size. A kilobyte is 1,024 bytes, each byte being a group of eight binary digits.
Flash memory offers an average speed of 1GB per second.