Flash memory holds whatever a device must remember when the power is off: its firmware, the bootloader, credentials, configuration and any telemetry queued while the network is down. Microcontrollers include some on-chip; larger needs are met with an external chip on an SPI bus, an eMMC part, or on Linux-class devices an SD card or SSD. NOR flash is common for code because it can be executed in place; NAND offers more capacity for data.
The property that catches designers out is wear. Each block tolerates a finite number of erase cycles, so a device that logs a sensor reading to flash every second can wear out its storage long before the product’s expected life. Mitigations include wear leveling in the file system, batching writes in RAM, and reserving a dedicated region for frequently changed data.
Part of the IoT Glossary. Spotted an error or a missing term? Tell us.