Understanding Flash Memory: Types, Architecture, and Storage System Design

Flash memory, encompassing NOR and NAND variants, serves as a fundamental non‑volatile storage technology, with NOR flash used for boot loaders and OS code due to its execute‑in‑place capability, while NAND flash offers high capacity and is the primary medium for modern SSDs.

The basic storage unit of NAND flash is the cell, which can store 1, 2, or 3 bits depending on the cell type: SLC (single‑level cell), MLC (multi‑level cell, including eMLC and cMLC), and TLC (triple‑level cell). Each cell’s programming (charging) and erasing (discharging) constitute a P/E cycle, and the endurance of a flash chip is determined by the allowable number of such cycles.

NAND flash devices are organized hierarchically: multiple dies form a device, each die contains planes, planes contain blocks (the smallest erasable unit), and blocks consist of pages (the smallest writable unit). SSDs integrate these flash chips with an embedded controller, cache memory, and firmware that manage tasks such as wear‑leveling, error‑correction, garbage collection, and data encryption.

To extend SSD lifespan, manufacturers employ over‑provisioning, providing extra spare blocks that replace worn‑out ones. Advanced storage systems also incorporate techniques like deduplication, compression, thin provisioning, and global FTL to reduce write amplification and improve space efficiency.

Performance optimization for flash‑based storage focuses on reducing latency and maximizing random IOPS, requiring high‑speed interfaces (e.g., PCIe, 16 Gbps Fibre Channel, 56 Gbps InfiniBand) and scalable architectures such as active‑active controllers and cache algorithms.

Emerging trends include the adoption of the NVMe protocol for standardized communication with PCIe SSDs, and 3D‑Flash technologies—memristor‑based storage, stacked horizontal NAND, and vertical NAND (V‑NAND)—which increase density and performance while addressing scaling challenges.