What's the reason behind Linux being so successful?

[HolmesSherlock]

Hi Community,

This is apparently a non-techie question I want to put in front, but in no way less astonishing it is. What I learnt that when Linus Torvalds wrote Linux kernel, he was a college student, possibly at the age of 21 years or so. Thereafter, it has got a huge success, both for commercially and non-commercially. I find a number of people out here who seems to be talented enough, having their own OS-es, doing difficult to more difficult development. But, has any of them got such a success ever? Which one? By whom? If not, then why not? What are the ingredients that made an OS by a 21-years old college student put into? What are we missing here?

[Jezze]

In the early days I think it was because the timing was perfect because there was no good alternatives really and people wanted to use a UNIX like operating system but could not find one without having to pay a lot of money. I know BSD existed back then but if memory serves me well it was not free as it is today and when it eventually became open it was already to late. If they had taken that decision earlier we might had been running BSD today. Minix was just a educational tool and most other operating system were closed or not good enough. The GNU people were busy developing HURD but that never really got anywhere so when Linux came many people who had been waiting for HURD quickly jumped onto Linux because it already worked out of the box and actually fulfilled most of people's wishes already - you have to give Linux credit for that. I think the GPL licence could have been a contributing factor but not as much as actually having something that works. Linux has since the beginning always had a very trusty userbase and once you make people accustomed to something it is a difficult to get them to convert to something else especially today when Linux has so much hardware support it is ridiculous.

The tragic side about Linux's success is that now it acts like a big fish that eats all the other fish in the pond so there is not much room left for any other operating system to compete for the mainstream audience and this has actually changed drastically in the last 10 years. It used to more competitors just a decade ago but they seem to all have vanished.

There are plenty of good alternatives, if you look for them, that on a technical level is more superior but because of the lack of drivers et.c. it just can't compete on the same terms. My favourite Plan9 is a good example of that. Haiku is another good example that even though it is pretty and user-friendly does not seem to generate much interest.

[HolmesSherlock]

In the early days I think it was because the timing was perfect because there was no good alternatives really and people wanted to use a UNIX like operating system but could not find one without having to pay a lot of money. I know BSD existed back then but if memory serves me well it was not free as it is today and when it eventually became open it was already to late. If they had taken that decision earlier we might had been running BSD today.

Just to reinforce your thoughts.
http://www.computerhope.com/history/unix.htm

Year Event
1957 Bell Labs found they needed an operating system for their computer center that at the time was running various batch jobs. The BESYS operating system was created at Bell Labs to deal with these needs.
1965 Bell Labs was adopting third generation computer equipment and decided to join forces with General Electric and MIT to create Multics (Multiplexed Information and Computing Service).
1969 By April 1969, AT&T made a decision to withdraw Multics and go with GECOS. When Multics was withdrawn Ken Thompson and Dennis Ritchie needed to rewrite an operating system in order to play space travel on another smaller machine (a DEC PDP-7 [Programmed Data Processor 4K memory for user programs). The result was a system that a punning colleague called UNICS (UNiplexed Information and Computing Service)--an 'emasculated Multics'.
1969 Summer 1969 Unix was developed.
1969 Linus Torvalds is born.
1971 First edition of Unix released 11/03/1971. The first edition of the "Unix PROGRAMMER'S MANUAL [by] K. Thompson [and] D. M. Ritchie." It includes over 60 commands like: b (compile B program); boot (reboot system); cat (concatenate files); chdir (change working directory); chmod (change access mode); chown (change owner); cp (copy file); ls (list directory contents); mv (move or rename file); roff (run off text); wc (get word count); who (who is one the system). The main thing missing was pipes.
1972 Second edition of Unix released December 06, 1972.
1972 Ritchie rewrote B and called the new language C.
1973 Unix had been installed on 16 sites (all within AT&T/Western Electric); it was publically unveiled at a conference in October.
1973 Third edition of Unix released February 1973
1973 Forth edition of Unix released November 1973
1974 Fifth edition of Unix released June 1974
1974 Thompson went to UC Berkeley to teach for a year, Bill Joy arrived as a new graduate student. Frustrated with ed, Joy developed a more featured editor em.
1975 Sixth edition of Unix released May 1975
1975 Bourne shell is introduced begins being added onto.
1977 1BSD released late 1977
1978 2BSD released mid 1978
1979 Seventh edition of Unix released January 1979
1979 3BSD released late 1979
1979 SCO founded by Doug and Larry Michels as Unix porting and consulting company.
1980 4.0BSD released October 1980
1982 SGI introduces IRIX.
1983 SCO delivers its first packaged Unix system called SCO XENIX System V for Intel 8086 and 8088 processor-based PCs.
1983 The GNU operating system is first announced by Richard Stallman September 27, 1983.
1984 Ultrix 1.0 was released.
1985 Eighth edition of Unix released February 1985
1985 The GNU manifesto is published in the March 1985 issue of Dr. Dobb's Journal. The GNU project starts a year and a half later.
1986 HP-UX 1.0 released.
1986 Ninth edition of Unix released September 1986
1987 Sun and AT&T lay the groundwork for business computing in the next decade with an alliance to develop Unix System V Release 4.
1988 HP-UX 2.0 released.
1988 HP-UX 3.0 released.
1989 SCO ships SCO Unix System V/386, the first volume commercial product licensed by AT&T to use the Unix System trademark.
1989 HP-UX 7.0 released.
1989 Tenth edition of Unix released October 1989
1990 AIX short for Advanced Interactive eXecutive was first entered into the market by IBM February 1990.
1991 Sun unveils Solaris 2 operating environment, specially tuned for symetric multiprocessing.
1991 Linux is introduced by Linus Torvalds, a student in Finland.
1991 HP-UX 8.0 released.
1991 BSD/386 ALPHA First code released to people outside BSDI 12/xx/1991
1992 HP-UX 9.0 released.
1993 NetBSD 0.8 released 04/20/1993
1993 FreeBSD 1.0 released December of 1993
1994 Red Hat Linux is introduced.
1994 Caldera, Inc was founded in 1994 by Ransom Love and Bryan Sparks.
1994 NetBSD 1.0 released 10/26/1994
1995 FreeBSD 2.0 released 01/xx/1995
1995 SCO acquires Unix Systems source technology business from Novell Corporation (which had acquired it from AT&T's Unix System Laboratories). SCO also acquires UnixWare 2 operating system from Novell.
1995 HP-UX 10.0 released.
1995 4.4 BSD Lite Release 2 the true final distribution from the CSRG 06/xx/1995
1996 KDE is started to be developed by Matthias Ettrich
1997 HP-UX 11.0 released.
1997 Caldera ships OpenLinux Standard 1.1 May 5, 1997, the second offering in Caldera's OpenLinux product line
1998 IRIX 6.5 the fifth generation of SGI Unix is released July 6, 1998.
1998 SCO delivers UnixWare 7 operating system.
1998 Sun Solaris 7 o

[Brendan]

Hi,

Just to reinforce your thoughts.
http://www.computerhope.com/history/unix.htm

This is missing a significant event:

1991 Linux is introduced by Linus Torvalds, a student in Finland.
1991 HP-UX 8.0 released.
1991 BSD/386 ALPHA First code released to people outside BSDI 12/xx/1991
1992 AT&T start copyright/trademark lawsuit against BSD, causing a significant number of developers to shift from BSD to Linux
1992 HP-UX 9.0 released.
1993 NetBSD 0.8 released 04/20/1993
1993 FreeBSD 1.0 released December of 1993

Of course once you've got more developers and more working drivers than other free *nix clones, you tend to stay in the lead.


Cheers,

Brendan

[Antti]

What are the ingredients that made an OS by a 21-years old college student put into? What are we missing here?

He wanted to have a good Unix(-like) system on his new PC (Minix was not sufficient). They were too expensive to be bought so it was easier to write it by hand.

The main thing we are missing nowadays is that the current operating systems are too good already. We really do not need to create a new one. When Linus created the first usable version of Linux, it actually was better than the alternatives (at least MS-DOS, the mainstream OS). Then he used the new system as his primary OS and did all the things with it. It was not just a hobby OS that is booted once in a while and the real work is done with the other "real" OS.

Nowadays, it is quite hard to make an OS that really is better for everyday use than the "real" OS. We may play with our own OS but we are very keen to boot, for example, Linux based OS or Windows when we want to do some real work.

[bluemoon]

Agree with Antti.

While nowadays most people can write an OS that beat DOS within one or two years, if you target to replace an modern OS it would take more than ten man-years of work.

[shikhin]

Hi,

Agree with Antti.

While nowadays most people can write an OS that beat DOS within one or two years, if you target to replace an modern OS it would take more than ten man-years of work.

Microsoft admit that it took roughly 90 man-years for the original Windows to be developed. Estimates for Internet Explorer 3 are at 10,000 man years. 10 man years? Understatement?

(reference: http://answers.google.com/answers/threadview?id=305678, most of the links seem dead)

Regards,
Shikhin

[Combuster]

Ten man-years if you don't plan on writing everything from scratch. Which is why unix clones here seem to have such success: you write a kernel and for the rest you have nothing to do but just copying all the existing work.

And 10k man-years on internet explorer sounds rather overrated, since it implies having about 3000 full-time employees on a single codebase. No wonder it went nowhere.

[bluemoon]

10 man years? Understatement?

it would take more than ten man-years of work. I just want to point out that the work is so huge that it is not easy to accomplish, compare to 1~2 man-years work for the original Linux.

[HolmesSherlock]

1992 AT&T start copyright/trademark lawsuit against BSD, causing a significant number of developers to shift from BSD to Linux

I understand that some 20 years back, developers did not have much options in front of them to work on, I mean not many OSes were there. So, a part of them, when forced to move to some other one, they chose Linux. Well and good. But, also we need to understand that Linux was far away from being a "mature" OS, it was basically a kernel. Even today, after hours of googling, I wonder much that how much of "what we call Linux" is actually "Linux". All the utilities, GRUB, GNU tool chain, GNOME/KDE desktop environments - are third party developments and not developed by Torvalds. Isn't it surprising that those hard-core devs readily picked up a kernel developed by a college student without developing their own? Torvalds, either did something excellent even being a college student or *somehow* he made them to work on his brainchild. Now history knows what the truth is.

[Kevin]

Linux was there, and it worked, and it allowed them to hack on it. What other kernel should they have picked that would have allowed them the same? The HURD isn't ready even today...

[bewing]

As one of the oldest coders here, I need to add one thing that was perfectly clear to me at the time -- and was widely discussed at the time. Linux may not have been a "mature" OS in the early 1990s, but that was not what anyone needed it for. You see, there was a new computer trend going on at that time. People were starting to hook their PCs together in these odd things called "local area networks", with this odd wiring called 10baseT that was replacing the old buggy coax, and it turned out you needed these special PCs called "servers". And you needed an OS to stick on these server things. You could try Windows 3.11 (for workgroups), but it crashed all the time. You could buy a copy of SCO, but it was really expensive.

Or you could use Linux. It was free (except for tech support from RedHat). And it had very good uptime compared with the others for servers. If you were a corporation building out a local network, then you needed to hire a couple guys to manage it anyway. And then the Web came along and made it even more important to have a network of servers. So Linux accidentally happened to fill a hardware need at just the time it came along.

So the analogy implies that once Android gets built out into a "mature" OS and diverges from its roots, it is probably going to be the next successful OS 20 years from now.

[HolmesSherlock]

Hope that people who are participating in this discussion is enjoying a lot. Not only building OSes, but digging into the success story of an open-source one poses much of importance.

While searching, I came across this page focusing on Unix/Linux history.

It says

In 1976-77 Thompson took a six-month sabbatical to teach as a visiting professor at the Computer Science Department at the University of California-Berkeley (UCB). He was of course teaching UNIX.

When Thompson returned to Bell Labs, students and professors at Berkeley continued to enhance UNIX. This led to Berkeley Software Distribution (BSD) V4.2

Wait! Was UNIX free(-ly distributable) at its early stage? Unless had it been so, how could after Ken returning to Bell Labs, UCB people continue working on it?

Also, there is an age-old announcement from "The Horse's Mouth" in the Minix group before the inception of Linux era:
https://groups.google.com/forum/?fromgr ... lNtH7RRrGA

Hello everybody out there using minix -
I'm doing a (free) operating system (just a hobby, won't be big and
professional like gnu) for 386(486) AT clones. This has been brewing
since april, and is starting to get ready. I'd like any feedback on
things people like/dislike in minix, as my OS resembles it somewhat
(same physical layout of the file-system (due to practical reasons)
among other things).

I've currently ported bash(1.08) and gcc(1.40), and things seem to work.
This implies that I'll get something practical within a few months, and
I'd like to know what features most people would want. Any suggestions
are welcome, but I won't promise I'll implement them

Linus ([email protected])

PS. Yes - it's free of any minix code, and it has a multi-threaded fs.
It is NOT protable (uses 386 task switching etc), and it probably never
will support anything other than AT-harddisks, as that's all I have .

[gravaera]

Yo:

I really, really think Brendan and Bewing got it pretty well down.

--Peace out,
gravaera

[Antti]

The HURD isn't ready even today...

Maybe HURD would have been ready much earlier (1994?) if Linux had not taken its place? There is/was no immediate need for it because Linux already exists/existed. If Linux did not exist, we would probably talk about GNU (operating system) here. It is a mystery for me whether it would be better or not. Maybe GNU (operating system) development would have been more under control and system-wide decisions of the development direction could have been made.

Moreover, the word kernel might be less used because there were no need to emphasize it being a "separate project" than the common userspace tools.

[...]

This was a very interesting viewpoint.