History and ecosystem of UNIX and Linux

Linux as you administer it is the result of three independent lineages that met in the early 1990s. Confusing them produces wrong mental models and wrong production decisions.

The three lineages

flowchart LR
  subgraph A[UNIX lineage]
    A1["Bell Labs (1969)"] --> A2["BSD (1977)"]
    A1 --> A3["System V (1983)"]
    A2 --> A4["macOS, FreeBSD, OpenBSD"]
    A3 --> A5["Solaris, AIX, HP-UX"]
  end

  subgraph G[GNU Project - 1983]
    G1["GNU C Library (glibc)"] --> G2["GNU coreutils"]
    G1 --> G3["GCC"]
    G1 --> G4["Bash"]
    G2 --> G5["GNU userland"]
  end

  subgraph K[Linux kernel - 1991]
    K1["Linux 0.01 (1991)"] --> K2["Linux 6.x (today)"]
  end

  K --> R[Linux distributions]
  G --> R
  A -.derived APIs.-> R

  R --> R1[Debian / Ubuntu]
  R --> R2[RHEL / Rocky / Alma]
  R --> R3[SUSE / openSUSE]
  R --> R4[Arch / Fedora / others]

1. UNIX

In 1969, Ken Thompson and Dennis Ritchie at Bell Labs wrote a small timesharing system on a PDP-7. They ported it to the PDP-11 in 1970 and rewrote it in C in 1973, which made UNIX portable to other hardware. Two academic forks — BSD at Berkeley and the System V line at AT&T — defined the modern API surface: fork(), execve(), open(), the shell, the init process, and the “everything is a file” philosophy.

The UNIX wars of the late 1980s produced POSIX (IEEE 1003.1), the standard that Linux, BSD, and macOS all aim to satisfy.

2. The GNU Project

In 1983 Richard Stallman launched the GNU Project to build a free UNIX replacement. By the early 1990s GNU had produced almost everything a UNIX-like operating system needs — a C compiler (GCC), a C library (glibc), coreutils, Bash, the GNU debugger, GNU Emacs — but the kernel was missing.

3. The Linux kernel

In 1991 Linus Torvalds, then a student in Helsinki, released a hobby kernel. It adopted the GNU userland, which had been waiting for exactly this piece. The combined system became “GNU/Linux”, though in common usage it is just “Linux”.

Kernel vs userland vs distribution

These three terms are not interchangeable:

A “Linux version” is almost always a distribution version, not a kernel version. Reporting a bug as “Linux is broken” without naming the distribution and the kernel package is a strong signal that the diagnostic work has not yet started.

$ cat /etc/os-release

NAME=Ubuntu
VERSION=24.04 LTS (Noble Numbat)
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME='Ubuntu 24.04 LTS'
VERSION_ID=24.04
VERSION_CODENAME=noble
UBUNTU_CODENAME=noble

$ uname -r

6.6.31-linuxkit

The two main families

Two commercial-grade families dominate the production landscape. Labs in this course target both and call out the differences.

Debian family

• Origin: Debian Project, founded 1993 by Ian Murdock.
• Package format: .deb, managed by apt / apt-get / dpkg.
• Init system: systemd (since Debian 8).
• Major derivatives: Ubuntu, Linux Mint, Pop!_OS.
• Release cadence: Ubuntu ships a non-LTS every 6 months and an LTS every 2 years (current: 24.04, supported until 2029). Debian stable ships every ~2 years.

RHEL family

• Origin: Red Hat Enterprise Linux, first released 2003.
• Package format: .rpm, managed by dnf / rpm.
• Init system: systemd (since RHEL 7).
• Major derivatives: Rocky Linux, AlmaLinux, Oracle Linux, CentOS Stream.
• Release cadence: Major RHEL versions roughly every 3-4 years; minor releases approximately every 6 months. RHEL 9 is the current major.

The two families differ in package names, repository conventions, default firewall tooling (ufw vs firewalld), and the presence of SELinux vs AppArmor. The kernel and GNU userland are largely identical, because both are upstream projects.

Open-source licensing

The legal model affects what you can and cannot do in production.

• GPLv2: The Linux kernel. Modifications to the kernel you distribute must also be GPLv2. This is why vendors who ship customised kernels publish the patches.
• LGPL: glibc and many system libraries. You can link against LGPL code from proprietary applications with some conditions.
• GPLv3: Some GNU tools (recent versions of coreutils, GNU grep). Distribution of modified versions requires source publication under GPLv3 terms; some downstream distributions deliberately stay on GPLv2 versions to avoid the patent grant language.
• MIT / BSD / Apache 2.0: Most userland tools from BSD-derived projects, plus the kernel build tools and many utilities.

Knowledge check