Binary Endianness: Big-Endian vs. Little-Endian in Multi-Byte Systems

When a value occupies more than one byte — a 32-bit integer, a 16-bit sensor reading, a 64-bit floating-point number — the hardware must decide which byte to place at the lowest memory address. Big-endian (BE) stores the most-significant byte (MSB) at the lowest address, mirroring the way humans write decimal numbers with the largest place value on the left. Little-endian (LE) stores the least-significant byte (LSB) at the lowest address. For example, the 32-bit hexadecimal value 0x44332211 would appear in memory as bytes [0x44, 0x33, 0x22, 0x11] on a big-endian machine and [0x11, 0x22, 0x33, 0x44] on a little-endian machine. The reason this matters is that byte ordering is invisible as long as you stay inside a single CPU architecture. The moment data crosses a boundary — written to a file, sent over a network, shared between an ARM Cortex-M microcontroller and an x86 server, or even accessed through a type-punning pointer cast in C — the two sides must agree on byte order or the values will be silently misinterpreted. A 16-bit temperature reading of 0x0100 (256 in decimal) becomes 0x0001 (1 in decimal) if the receiver assumes the opposite endianness. This class of bug is notoriously hard to catch because the program does not crash; it simply produces wrong numbers. Historically, Motorola 68000 processors and most RISC architectures (SPARC, PowerPC in default mode) used big-endian ordering, while Intel x86 and x86-64 processors — and therefore the vast majority of desktop PCs and servers today — use little-endian. Modern ARM processors are bi-endian (they can be configured either way), but in practice ARMv7 and ARMv8 systems almost universally run in little-endian mode (Android, iOS, Linux on ARM). Network protocols, however, standardized on big-endian byte order decades ago — hence the term 'network byte order' is synonymous with big-endian. The POSIX functions htons(), htonl(), ntohs(), and ntohl() exist precisely to convert between host byte order and network byte order. In embedded systems, endianness surfaces constantly: when parsing binary sensor data over SPI or I²C, when reading registers from a peripheral whose datasheet specifies a particular byte order, when writing firmware that must serialize structs into flash memory or transmit them over CAN bus. In data serialization formats, Protocol Buffers and MessagePack define their own wire byte order (varint encoding and big-endian respectively), while formats like TIFF include an endianness marker in the file header ('II' for little-endian, 'MM' for big-endian). Understanding endianness is not about memorizing which CPUs use which order — it is about recognizing every point in your system where bytes cross a trust boundary and ensuring an explicit, documented conversion happens there.