In 1991, I was already an experienced Unix sysadmin and writer. I'm sure I saw Linus Torvalds's famous Usenet message: "I'm doing a (free) operating system (just a hobby, won't be big and professional like gnu) for 386(486) AT clones," and I paid it no mind. Many people said similar things and little came of it. This time, it would be different.
Torvalds was a brilliant programmer with a gift for getting others to work with him. This, combined with the use of the GNU General Public License version 2 (GPLv2), made Linux the operating system of choice for everything from supercomputers to smartphones. It didn't look that way at the start.
Ted T'so was the first North American Linux kernel developer. He started working with Linux in September of 1991 with the 0.09 kernel. Back in those days, the only way you could download Linux was over a slow connection from a Finnish FTP server. So, T'so set up and hosted the first North American Linux ftp server: his personal desktop.
Today he works at Google. He remembers that in Linux's early days one of the reasons why Linux was successful was because of its "stone soup approach. Everyone had their own interests and worked together to make sure their project or problems got worked on".
James Bottomley, an IBM distinguished engineer and Linux programmer, got his start in Linux. In 1992 he was working on a computational physics PhD at Cambridge. About that time, he said:
In those days Pentium PCs were 1/10 the cost of the usual Sun SPARC (or Alpha or any other computer system we evaluated) and the new Pentium Floating Point Unit promised equal performance in computational physics. We required Unix systems to integrate with the current setup, so I persuaded the department to invest in 3 Pentium systems running Linux instead of buying another expensive Unix system.
He wasn't the only pioneer who came to Linux because of cost. Gerald Pfeifer, SUSE's senior director of products and technology programs, said:
I had been using classic Unix systems (DEC, Solaris,...) for years, before Linux existed (let alone was more widely known). Working on my PhD thesis and as an assistant professor at Vienna University of Technology we had Solaris workstations, which turned out to be both expensive and slow, and started supplementing those by industry standard hardware (aka x86) first running FreeBSD and then later Linux.
Today, we've forgotten about it, but in the 90s one reason why Linux was so attractive was it much cheaper than high-end Unix systems. There were Unix systems for x86 processors, such as NeXTStep, MacOS's ancestor, SCO OpenServer, and Dell SVR4 Unix -- yes, Dell used to have its own Unix -- but relatively few people used them. The BSD Unixes, which came before Linux, had a measure of support, but Linux quickly became far more popular.
Bottomley is still surprised that Linux "eclipsed all other open-source operating systems and just stayed there". He continued:
In 1992 the USL lawsuit was just wrapping up, so my initial target for the Pentium systems was BSD. It was just an accident that AT&T dragged the suit out too far and Linux became the first viable free OS for them, so I tried it instead. It worked well enough then and 386BSD never really took off, so we stayed with it once we'd deployed. By the time the other x86 BSDs finally emerged it was too little too late. There were many ways Linux could have tripped up and then have the BSDs roll over it, but it didn't. All the fundamental decisions made in the early days were basically correct. You've no idea how unusual this last thing is in the industry.
One reason why Bottomley thinks this happened is because of Torvalds' leadership style. "There's a very under appreciated fact about management: you don't have to make the correct decisions up front (how can you, you usually don't have enough information) you just have to make an swift but correctable decision and allow crowd-sourcing to work for the correction."
Pfeifer's take on Torvald's management style, who he met in 1994, is Torvalds' "focus on only allowing the 'right' and 'best' code in while still motivating more people to contribute was IMHO a key aspect of his character and his leadership."
Ransom Love, former CEO of Caldera Linux, in its day a major Red Hat rival, was one of the first people who saw Linux's business potential. Love explained, "I was working as a Product Manager for Novell on the future versions of NetWare and was working with a number of the engineers experimenting with micro kernels and different technologies. One of them was Bryan Sparks. He introduced me to Linux. At the time, there were just a few people using it. Slackware was just starting to come on as a leader in the distribution area. In talking to Bryan, we felt we could create a desktop that showcased all the networking features and give it away for free with NetWare."
Unfortunately for Novell, which would eventually end up buying SUSE Linux, 10 years later, Novell wasn't ready for Linux. So, Sparks and Love left Novell to form Caldera.
While Caldera would become entangled in the SCO Linux lawsuit fiasco, Love left before that disaster started to unfold. He continued to be a Linux and open-source booster. Love is still surprised at "the quantity, quality, and capabilities of the people who were drawn to it and open source. The concept of sharing your ideas and technology openly is still fundamentally changing society and bringing wave after wave of technology innovation".
Others would come to Linux in the mid 90s. Clement Lefebvre, founder of Mint Linux, "started with Slackware in the 90s. Floppies were going around the university. We were using AIX there, and so the prospect to run a Unix-like at home seduced many of us. Add open source to that and it was love at first sight. I stuck with Slackware for about 10 years after that, and I've very fond memories of it".
Still, others had more pragmatic reasons. Matthew Miller, the Fedora Project Leader, explained, "A friend and I started an ISP in the mid 90s. We provided the first-ever Internet connectivity to the Amish stronghold of LaGrange, Indiana. (I am not making this up!) Our core infrastructure was initially Windows NT, and we had one server which crashed every night, no matter what we tried. So, one late evening (or early morning), we decided that it couldn't be any worse with this Linux thing we'd heard about."
He continued, "So, I ordered one of those five-CD distro sets by mail... the first disc was Red Hat Linux, but that didn't boot for some reason. Slackware was next, and that worked, so Slackware it was. It's funny -- my whole life probably would have been different if Debian had been sorted differently. Anyway, I don't think it was more than a few weeks before we decided that the other servers needed to be converted too. Switched my workstation over too, and I've not used another OS at any serious level ever since."
He's not the only one. While Linux hasn't conquered the desktop, it's omnipresent. It's in your Android phone, it powers most of the websites you visit, and it runs your Wi-Fi routers, TVs, and DVRs.