Google and Sun
This CNET article discusses the decline and failure of Sun and success of Google based on this central idea:
I would disagree with this. I think Sun’s failure goes back further and gave birth to the child that ended up killing it – Linux.
.
Back when I was a just starting off in software engineering, our product line ran on Apollo’s DomainOS – a great operating system. It was dying out and we were porting to Sun’s Solaris operating system. I was new to Unix and soon developed an appreciation for this complex and powerful environment. Wanting to learn more, I started looking for a version of Unix to run on my home PC – so I could hack on it to my heart’s content without worrying about messing up one my company’s $15K workstations. First stop was Solaris x86 – a version of Solaris that could run on a home PC. As I recall, it was something like $300. Someone told me about Linux and I ordered a CD from TransAmeritech LinuxWare. It was cheap – certainly less thatn $50 and maybe down around $10 or $20 dollars.
.
After a minimum of fuss, I had it installed and I can still remember the first time my home PC booted up with it’s very own Unix-like system … Linux. I’ve been hooked ever since. It came with a compiler, several desktop environments, editors, games, and applications galore. In many ways it was far more usable as a desktop than my $15K Solaris machine. Sure the same applications were availabe but you had to download, compile, and install them yourself.
.
In 1991 when university student Linux Torvalds wanted a unix for his PC, he wrote Linux. Me and millions of other Unix users embraced it because of Sun’s failure to make a free version of Solaris available to home users. The opensource nature of Linux fostered a huge ecosystem that led to rapid development and better distributions and applications. Meanwhile Sun plodded along with their off-again on-again relationship with Solaris x86. Eventually Sun saw the light. They opensourced Solaris and started shipping GNU stalwarts like Perl and Gnome with the operating system — but it was too little to late. Us early Linux adopters were now sitting on top of a mature and totally free powerhouse. It ran on cheap hardware and we were in position to influence our companies to adopt Linux internally. I was one of people pushing to migrate a move to Linux within my company – which we did leaving Solaris behind.
.
That’s my story. Sun didn’t fail because they didn’t embrace Linux on x86 – they failed because they tried to keep all their technological marbles – creating an environment that encouraged Linux to thrive. Me and millions of others never would have touched Linux if Sun had a free or cheap version of Solaris or SunOS for our PC’s. It’s too bad. Sun hardware and OS are rock solid and crazy scalable. The company has been a good opensource citizen(OpenOffice, Mysql), innovative(Java, NFS, ZFX, Dtrace) and I’m sad to see them go.
Themocracy
From what you describe Microsoft products should not exist. The only thing Microsoft “gives away” is trialware. And it’s dead after a short period of time.
I started using Solaris about the same time as you. I installed hundreds of systems running Solaris, HP-UX, AIX and other versions of Unix. Now I fix them and install Linux (RedHat, SUSE, Ubuntu and others). I see the downfall of Solaris as a marketing problem. I remember Word Perfect, dBase, DEC and all the rest of the companies that have not made it because the could not sell the product and would not change to fit the new markets. Sun just followed their steps.
Good products will never triumph over good marketing.
@Bob
A fair observation.
I see Microsoft somewhat differently. By becoming the OS vendor for IBM personal computers, they were given a defacto monopoly in that space. IBM in what surely must be one of the most terrible decisions in the history of business, saw profit from PC’s coming almost solely from hardware – not terribly unlike Sun and their proprietary Sparc processors. While M$ didn’t give the OS away, the vast majority of people never paid for it since it came pre-installed and the markup to cover the OS cost was relatively low given the multi-thousand dollar pricetag on PCs when they first came out. Even today, most people don’t buy Windows – it’s cost is largely invisible and bundled into new PCs.
With their monopoly and shady business practices, they hung on to their market dominance until the day when the OS wasn’t as embarrassingly shabby has the pre-NT/XP versions (Win3.x, Win95,98,ME). Now inertia is on their side but with Linux and Apple both producing viable alternatives, and Google waiting in the wings with Android, they’re not the 800 pound gorilla they once were. Maybe just s 700 pound one.
Microsofts business shenanigans make Machiavelli look like a chump.
SCO had a free Unix for a brief period, single user, but it included the development system, (even C++) and even Arcserve for backup – but it was far too late, because Linux was already ascendant.
Imagine if Sun and SCO had even just student versions in 1991 – Linus never would have bothered and we would have had to wait for Mac OS X to get an affordable Unix system!
My point exactly.
Google (GOOG) is getting beat up after hours. Lousy quarter. But I love it with my meager shares. When will they finally split? Sun (SUN) is a loser. Have to be nuts to spend a dime on it. Want to toss your money away on a risk? Try Ambec (ABK). Mr. Softy (MSFT) is just great. They have some real smarts to generate the returns they do. Top of the line in – wish I had ground floor entry. They have enough free cash to fund HC. Red Hat (RHT) has been flatline for a long time. Rambus (RMBS) is a better risk. Then again, avoid all the thinking and go Power Shares (QQQQ). I have no idea what SCO is but if it is Pro Shares Oil Short then you are dead! I love it when the shorts get screwed. Apple (APPL) is a winner. That baby has gone up 130% in about 8 months.
Linux was never intended to be a real OS. It was an academic exercise that caught on because of the nascent OSS movement at GNU. there was an OS called MINIX that was the jumping off point for Torvalds. You should read his Usenet posts on the subject.
@Ebenezer Attaquin
I have read those old usenet posts – interesting stuff. We also studied a bit of the Minix source code in one of my CompSci classes in high school – specifically the fork system call.
.
On a related topic since you mentioned usenet posts, you’ve probably seen the classic flamefest between Torvalds and Tannenbaum – great stuff.
Solaris is a dream compared to the old SunOS 3 that I used to support on Sparcstations and the weird 386i. But nothing was worse than SCO Xenix. I supported a system consisting of a Compaq 386/16 with a bunch of Wyse terminals and printers. Word processing and an inventory app written for Foxbase really taxed the system. Now I run a 64-bit dual core 4 gig terabyte Slackware 64 monster to surf and blog. My how things have changed.
Slackware. Nice system. I ran slackware for a number of years starting when you had to install with floppies. I went to Slackware right after TransAmeritech LinuxWare and really enjoyed it. Hardware detection in those days was painful and it eventually drove me to RedHat. This was RH4 or 5 which had a gawdawful FVWM desktop. I think from there I went to Mandrake which I loved. It had all the nice stuff I use to go get for myself(xemacs for instance). Let’s see from Mandrake I think I went to Fedora for a number of years and started playing with Debian based systems on the side starting with Mepis. I’m running mostly Ubuntu these days though in work we run CentOS so I can still keep up on RPM based systems.
Slack has made huge strides in hardware detection. You wouldn’t recognize it. Some weird sound cards may require manual interverntion. I’m stuck with slack because I once tried SuSe, and it removed all the manual changes I made. Ohhh… that was a feature!
SCO Xenix.. boy that takes me back. Tandy 6000′s before that, SunOS..
And there was, very briefly, a Microsoft Xenix. I helped a customer evaluate that once.. he stayed with the Tandy Xenix and then switched to SCO later. I upgraded many a SCO Xenix to SCO Openserver, did a little with Unixware, a lot of Solaris.
And now: amazingly enough, I still hear from SCO Unix folks. Most have moved off to Linux or Windows, but it’s really surprising how many are left. I was recently involved in a two year project to transition a SCO system to Windows. The Windows programmers were less than brilliant (something I kept mentioning to the client) and eventually it all blew up and the project was canceled, so he’s back on SCO Unix for at least a little while longer.
Just five tears ago I actually had a call from someone running a Tandy 6000 Xenix system – how that has stayed alive all these years has to be a minor miracle. Yes, they were moving to something else
Speaking of relics — I once used a Data General “minicomputer” that was almost as big as a desk. It ran a coordinate geometry program (used by surveyors) that was loaded from paper tape. The I/O was a teletype console! then there was the AT&T 3B2…
OK – you crusty graybeards have out-olded me. I cut my professional teeth on the Data General Nova 3 computer.
I take back what I said about Slack and Floppies. My first SlackWare came on CD in a book by Patrick Volkerding. I had 386 with a Mitsui cdrom drive – which wasn’t recognized during cd install. I had to make a boot floppy off images on the disk – which contained the necessary drivers. I also had a lot of trouble with the sound card on this machine – which wasn’t all bad since part of the goal was to learn unix/linux. While I haven’t built a kernel in a couple of years, it’s nice to know that I can.
There is actually a non museum site in India that still operates a punch card system – at least they did a few years ago. Saw an article on it in USA Today. Had an IBM 1401! That’s a 50 year old machine. Even had a few 407′s and the usually sundry equipment such as collators, key punch, sorters and the world famous jam machine the 552 interpreter. Where did they get the punch cards? I imagine it is like showing up at the airport for a cross country flight and your triple 7 was been replaced by a DC-3.
I used to build kernels on a 386 just to reduce the memory footprint. First one took all night to compile.
Later got a 486/25 motherboard when they were the latest thing. Installed the latest Slack version and wow! I did some work in a building where somebody was switching from a Unix system to Novell. They were carting perfectly good terminals and long serial cables to a dumpster. I managed to fit 3 in my trunk. I put the 486 in a basement space, and the terminals in the kitchen, my bedroom, and in the spare room where I had my ham radio equipment. Most fun I ever had.
Tried to cross-compile a kernel for the pair of Sun 3/50 that I had, but no dice. They are in a landfill now.