Evolution and Concepts of Microkernel Architecture

UNIX, created in 1971 at Bell Labs, became the classic general‑purpose operating system and introduced the macrokernel design where most system services reside in the kernel.

As UNIX kernels grew in functionality, maintainability and security challenges led researchers to explore microkernel designs that move many services to user space while keeping only minimal kernel functions such as process scheduling and inter‑process communication (IPC).

The first generation microkernel, Mach, demonstrated feasibility by using IPC and virtual memory to relocate services outside the kernel, but early versions suffered performance penalties and legal issues with incorporated BSD code.

The second generation, represented by L3/L4 and the Neutrino kernel of QNX, focused on dramatically simplifying IPC, using register‑based messaging and limiting data copying, achieving up to 20× faster IPC and finding commercial success in embedded and high‑reliability domains.

The third generation, including seL4, emphasized security by introducing capability‑based IPC, where processes must hold unforgeable tokens to request services, and achieved formal verification of the kernel’s correctness.

Other modern microkernels such as Google’s Zircon (used in Fuchsia) and Minix illustrate continued interest, offering modularity, fault isolation, and self‑healing services.

Advantages of microkernels include modular system services, higher portability, improved kernel security, ability to run multiple service sets concurrently, stable interfaces, reduced licensing constraints, suitability for formal verification, real‑time guarantees, and efficient multi‑processor designs.

Disadvantages involve extra overhead from IPC and context switches, increased latency for frequently used services, higher memory copying between modules, complexity in designing communication interfaces, and poorer cache locality due to separated address spaces.