Let's forget the last few years ever happened — the last five, at least. Possibly 10. In the 1980s Sun Microsystems was on fire. Founded in 1982, Sun raked in so much money that it broke the psychologically important $1bn sales barrier in six years. It took Microsoft 15 years to hit $1bn — six if your starting point is the date …
I do have the honour of having left a broken Sparc 10 on a tube train once! I was taking the system home to see if I could get it back up and running again. I got on a Piccadilly line train, got off, got home and realised I'd left the machine behind!
I suppose some daft sod pinched it and thought it was PC, when it wouldn't work chucked it in the local landfill!
"This was step one in making Solaris more accessible to the community. Step two, putting Solaris on x86"
In our shop, we where running Solaris x86 long time before Sun began to talk about open sourcing it. And by the way we eventually switched to Linux because it gave us much better performance on the same hardware, for our workload (that was a mail system frontend).
Mr. McNealy must have lived in a different Universe...
Yes. McNealy did live in a different universe. The problem with Linux wasn't that it was Free Software but that it was Unix on cheap x86 hardware. While Sun did have an early version of Solaris on x86, it was poorly supported. It was treated like an ugly red headed stepchild by both Sun and it's 3rd parties (like Oracle).
I personally started using Linux not because it was "free" or because it was "libre" but because it actually ran on my hardware. I would have actually been willing to spend the asking price for Solaris x86 back in the day if only it supported normal consumer grade PC hardware.
Sun tried to ignore the PC in favor of it's higher margin server business.
THIS is why Linux had the opportunity to sneak in and undercut Sun's business.
Sun's approach to x86 hardware ultimately did them in, not their approach to Free Software.
Just a bloody excellent article, really made my morning, have a bit of a lump in my throat to be honest as I am Ex-Sun myself.
While McNealy was a great former captain of the IT industry, I think Ellison has got it completely 100% spot on, he also has aggression I think is required to push Sun's technolgies and isn't afraid of playing a little dirty (raising core prices on Itaniums for one...)
Here's one Solaris guy who is quite a happy chappy at how things turned out, could have been a hell of a lot worse....
Not happy how it turned out
I'm glad you're happy how it turned out, because not many of us who are still at Oracle are enjoying it at all.
I'd certainly rather Sun had gone down fighting than watch everything I've worked on for the past decade be systematically trashed by Larry.
as an ex employee
When i first started at Sun, i was shocked by the lack of technical detail available to support staff. Technical documents were impossible to obtain and the ones that did exist were full of mistakes. Partner service companies were forced into throwing spares at machines to try and fix them.
Add to that the horrendous problems with the first versions of System V. Sparc Centres were a pain to repair,,,,often the fault had to be guessed at,,,,storage arrays were woeful to say the least. And sun spent millions trying to buy a storage business and failed every time.
Sun used off the shelf parts as you rightly pointed out, thrown together to make them cheap and a pain in the ass to fix.
The "back room" boys jealously guarded the IP, in case the bad news leaked to customers and when you add the "jobs for the boys" culture that was prevalent in many departments, you had a disaster waiting to happen.
But thats only part of the story because Sun could never decide whether it was a software company or a hardware company or both and ended up giving away the software which was the best bit of the business.
IBM knew what they were doing with Java and it became almost impossible for Sun to sell any of their development tools.
And who remembers N1 ? What a pile that was.
So there you have it, Sun spent too long at the party getting pissed on cheap wine and bragging about how MS was the Heroin dealer of the world.
Mcnealy was a great visionary but he created a monster that ran out of control.
I've been quite harsh on McNealy, but that was an interesting read. Obviously he didn't have that much control near the end. I did admire his empathy for his workers - that isn't BS on his part, Sun really did hold off on sacking people for as long as they could. Wall Street hated them for it though and it may have cost them in the end. You feel there is a human, empathetic side to McNealy whereas with Ellison it's all greenbacks and bravado.
Legends: McNealy, Bechtolsheim, Joy and Khosla.
All above are legends...Mr Ellison, now you hold the reins so-to-speak: one suggestion, please can you tell your marketing people to put the Sun logo letters back into italics (1990 version) as it should be ;-). Thank you.
PS - Keep up with the Iron Man sponsorship as well, good idea.
Hätte, Wäre, Wenn Wir Damals Doch, - LAME EXCUSES !
If this were a great man, he would simply state that he built a great business and then failed to see the competition and then failed the whole business.
He was the CEO and Chairman, for Spaghettimonster's sake, and he is going to tell us that "some managers" sabotaged Solaris on x86 ? What's the term for that ? Maybe "Weasel" ? Certainly not a "wild weasel", more a "geriatric, overweight weasel".
He first ignored NT (the Solaris GUI could never ever compete on Ergonomics) and then he ignored the Economic advantage of both x86 and Linux.
Now he claims Oracle got a bargain. Muhahaha. They bought a 5 billion dollar millstone for their neck. Larry should have bought himself a 100 million necklace in Paris, that would have been a much wiser decision.
At the time Oracle bough SUN, it was bled white. The processor slow as a slug, the Java stuff economically on the decline, the OS not better than Linux. From SUN to Millstone.
Sad end to a good company
I did some sysadmin on a network of Sun workstations for a few years during the nineties. These were really great machines able to blow away what you could do on the PCs of the day. While they could have done more with X86 by building and selling their own, the problems were competing Sparc hardware within the same company and X86 becoming a low margin commodity box shifting business.
The only way Sun could have squared the circle would have been to become primarily a services and support company, the way IBM and Red Hat did. That would have led to internal tensions, due to hardware divisions not wanting a service division to recommend non Sun hardware. To have survived, senior management would have had to have challenged departmental empire building from destroying the viability of other divisions; the services division should have been tasked with developing and providing world-class services and not with acting as a sales support arm of the Sparc hardware business.
@copsewood: In Short: McNealy Should Have Done His Work
...as a CEO. Not a Weasel.
Another failing of Suns..
.. was still trying to sell sparc workstations for 20K when the equivalent performance from an x86 PC would have cost 2K. This didn't matter so much when the only real OS competiion on x86 was Windows , but when Linux came to the fore that was the death knell for Solaris on the hi end desktop. And then slowly Linux started pushing out Solaris on the server side for the same reasons.
", therefore Oracle and not the community will run them." therefore the community will stop using them and use something else instead.
Customers and open source community overlap, not always
You assume community == all customers in the world. That's fair, since most people have been raised on open source software but it doesn't reflect the reality that enterprises are BUSINESSES that need to make money and buy whatever makes more sense to their objectives. If these business that are part of the "community" and have philosophic reasons to prefer "open source" over anything else (even if it's worse/better)... just look at where Sun is now. Good luck.
There is a way, Mr Coward, of developing and supporting free and open source without flushing your business down the toilet.
After all open source is about market economics, not monopolies. Anyone can make vast money from a monopoly. It takes a real capitalist to see a free market and make a go of it.
All praise Red Hat. etc etc.
Good article. Mr. McNealy, we salute you. Sun was a company that everyone liked - shareholders, employees, customers and users, and that is the best capitalism. I agree with McNealy's assessment of why Sun failed - they open sourced Solaris to aggressively, and too late to survive the 2008 crash.
Solaris is undoubtedly the best operating system today, in technical terms. Linux is catching up, and will close the gap sooner or later. Windows is years behind.
So, McNealy "believes in patents" but spent "years and millions of dollars to engineer out patents held by various patent holders"? I guess he's yet another person perpetuating the myth that "patents are great" while nursing the view that "our patents are great, everyone else's are garbage", thus revealing the selfish, anti-competitive monopolist nature of the patent enthusiasts and the patent system in general.
Then, according to McNealy, Sun "open sourced too much", but then insists that Solaris would have been in Linux's position had they released it as open source software in the first place. One might agree with the latter - as the 1990s progressed, people weren't going to mess around with gold-plated operating systems for expensive kit when everything else is running free-to-acquire and open stuff for commodity kit - but he can't then insist that opening up Java was a mistake when the credibility of Java beyond corporate "nobody got fired for specifying Java" shops was sliding faster than Sun's share price, much for the same reasons that Solaris was being passed over by people looking to adopt, learn, adapt and deploy stuff. (And yes, Solaris was available for x86 pretty early on, and people regarded it as a pretty poor substitute even for the nascent free-and-open stuff.)
What destroyed Sun was not the "open source hippy culture" that people like to pretend existed. It was the milking of the unsustainable dot-com era markets (targeting the stupid rich is a great idea only while the stupid remain rich, Scott!), the proprietary instincts and feet-dragging (undoubtedly, proprietary Solaris snobs probably pooh-poohed the idea of OpenSolaris, and then they released it under their own special licence rather than a widely used one), the inability to see how the Java monoculture had gutted the ecosystem (pandering to corporate shops with big middleware - which Sun never really succeeded with unlike, say, Red Hat - and, for a long time, having Windows as a primary platform but not Linux), and the inability to construct working communities around their much-trumpeted open source efforts (by wanting control over how everyone contributed to or even, in some cases, used the software).
Sun contributed a lot to the community, but the company's downfall was not due to any notion of excessive generosity. The finger of blame points right back at the company's management.
'So, McNealy "believes in patents" but spent "years and millions of dollars to engineer out patents held by various patent holders"?'
Yes. There's nothing contradictory there. If you believe in patents, and want to use a patented technology, you have two options: license it, or build your own equivalent that doesn't infringe the details of the patent. McNealy chose the latter. With open source, you often have to choose the latter, as many patent licenses include restrictions on revealing the methodology.
"as many patent licenses include restrictions on revealing the methodology"
Run that by me again.
It's not called a "patent" for nothing. It may be obfuscated and engineered to trap as many unwary souls as possible to generate rent, but there are no _restrictions_ on revealing anything. It's not trade secrecy.
Also, good luck on choosing the way of trying "not to infringe" when you don't even know whether you might.
McNeedy has obviously spent so many years spreading the Sunshine he's fallen for his own fairytale, hook, line and stinker of a lie. It always amuses me when Sunshiners insist that the only reason Sun failed on the desktop was because M$ "forced" Dell and co into making their PCs the way M$ wanted, completely ignoring the fact that the PC vendors happilly went along with it. Why? Because they stood to make massively more money that way because it was what the customers wanted, and Slowaris on the desktop was not. Windows was simply a far superior option, period! If it wasn't so, Dell and M$ would have whithered away and Sun would now be ruling the roost and would be the one in the position of being able to dictate licensing terms for SPARC and Slowaris to desktop vendors. Sun even gave away the SPARC designs for free and people still ignored it, instead reverse-engineering Intel's x86 design.
And McNeedy's hilarious insistance that it was only becuase Sun didn't open-source Slowaris sooner that Linux came about is just pure hogwash, completely ignoring all the other UNIX variants and "free" software work that happened both before Linus and before Slowaris. He wants to believe all Sun's custoemrs loved him and Sun? Well, as a former Sun customer all I can say is a string of four-lettered rejections of that line! Sun went from being a company I respected to being a complete pain in the rear in the twilight of McNeedy's reign.
TBH, I think Larry would simply laugh at the idea of McNeedy telling him how to run a business or work with the OSS community. IBM and hp diversified, Sun stagnated and died. AIX and hp-ux live on despite or probably because IBM and hp embraced Linux, whilst Sun flip-flopped between pretending to like the Penguin and openly hating it, cursing Intel and then having to relie on x64.
I can't believe that any tech webiste gives McNeedy the time to carry on spreading such unrepentant bilge. He may have helped build up (some would say just got lucky in more than one case) one of Silicon Valley's big names, but he also destroyed it through his own shortsighted ego. Sure, Ponytail didn't help, but the causes of Sun's death were the fatal mistakes repeatedly made against all the industry and market trends by Scott McNealy long before Ponytail got the job.
History lesson is a bit off
Under "History lessons" you postulate that "History might have been different had Sun open sourced Solaris sooner ...", following on from different recollections of Schwartz;
"In a 2003 interview ... Schwartz told eWeek that Sun paid AT&T to get rights equivalent to ownership ... We double-checked this with McNealy, and he is positive: there was an agreement with Novell in 1994. This would make more sense, given it was Novell that retained the Unix trademark"
Over at Groklaw the news picks editor noticed this article and Pamela Jones tartly pointed out that Schwartz and McNealy's are both right as far as it goes but both are confused. There was legally no choice here. There was a 1994 Novell deal but it merely removed royalty obligations. It still had confidentiality restrictions which did not permit Sun to open-source Solaris code. As Ms Jones puts it,
"If you read the appellate court's ruling in the first SCO v. Novell appeal, you'll find out precisely what happened. In 1994, Sun paid $83 million for buyout rights. But they still didn't have the right to open source the code. They got that right from the amendment to the 1994 deal in 2003, paying $9 million more".
The appeal in question is Case 08-4217 before Judge McConnell of the Tenth Circuit Court of Appeals. On page 50 it says, as Ms Jones quotes:
"After entering into its 2003 agreement with SCO, Sun released an opensource version of Solaris that would have been barred under the 1994 agreement".
Poor McNealy. He couldn't have open-sourced it earlier without paying a whole swag of money and seems not to have realized that. How many times did he call Time Out on an exec who told him so?
did his best. But failed. In my big famous finance company, there was policy from highest management to not buy Sun stuff. It didnt matter that Sun Niagara T2 hold several world records back then, for a fraction of the price of what similar vendors offered.
For instance, you needed six IBM POWER servers to match one single Sun T5440 in SIEBEL v8 benchmarks. Each POWER server costed $413.000 and the T5440 costed $76.000.
Now we have policy to not buy HP stuff. I dont get it. It seems that some companie's executive manegement share a black list "do not buy this company" or "buy this company". I talked to others, and they confirm that Sun was black listed. Even though Sun had cheaper and faster hardware. Who makes these black list?
*MOST* of the time, these strategic decisions are done on a golf course mate, logical choices rarely come into it.
Open core ?? WTF ??
Oh dear, dear, dear.
McNealy spouts the "Open Core" line, a description that's only been coined in the last couple of years max.
"Open Core" is just like all the M$ products and ideas with "Open" preceding their name - you know, like OOXML - i.e. the word "open" is redefined as "closed" aka "proprietary" in this usage of the word.
Maybe English will change its usage like it has since the KJV bible - you know, Paul was "previously let from going to Corinth" - back then "let" meant "prevent", now it means the exact opposite. "Open" is rapidly inverting its meaning too :-)
As another poster said, the license Sun used to release "open" Slowlaris was widely condemned by the community, and IIRC had to be modified to be meet the OSI's definition of Open Source.
Ever read the licenses in question, or do you just take RMS' rants as gospel?
CDDL is considered Open Source by ISO, even FSF acknowladges that it is a "free software license" (may I point out that the FSF is the FIRST group to always point out the difference between "free software" and "open source software"). It's really too bad the "Free" sotware foundation couldn't come up with a more free license then the GPL.
As a License it is MORE permissive then the GPL v2, and includes patent-protections (something the GPL v2 does not). The FSF's only two complaints with the CDDL are that it is incompatible with the more restrictive GPL v2 (which is true of any more-permissive "share-alike" license) and that it is "Also unfortunate in the CDDL is its use of the term “intellectual property”".
Frankly the FSF's need to condemn anything using the phrase "intellectual property" is childish, and the list of licenses that are incompatible with the GPL v2 include such closed licenses as the GPL v3!
486i vs sparc
Their fate was sealed when they chose their first gen SPARC over the better performing 486i prototypes.
Oh let's not get into this one
Single vs. multiple thread performance, server vs. desktop, yadda yadda
@Oh let's not get into this one
I think he is right. Deutsche Börse now moves fully to Linux from Solaris and OpenVMS. They are one of the largest derivatives trading platforms by volume. That anecdote is telling in my opinion.
Not to *fix* the Mbuf overflow bug that would panic a sun but to accept and process a fix staff at then $work wrote for the high end sun's by reverse engineering the relevant lib and recomiling with a fix. When the problem, test and fix was passed to Sun (no charge) they asked *us* for 10K (UKP) for "processing" costs. We declined and shipped our fix to customers direct.
Sun never properly fixed this glaring hole.
There is a real opportunity for Solaris on x86. If you turn down the volume on the Linux fanboys and listen to engineers at big e-commerce sites, you will hear non stop horror stories about the instability and general bloated chaos of Linux. They desperately want an operating system with the stability of MS Server but without the complex and expensive licensing. They want to find engineers who understand programing and testing and don't just google for a patch when Linux crashes and burns on their servers.
But Oracle may see more profit in being less open with Solaris. They are already making out like bandits with their proprietary database system, which you pretty much have to use with Linux on anything major. If you ever wonder why there is almost a religlous fatwa by the MySQL community against TPC-C benchmarks -- if you manage to find an illicit MySQL test -- you see how badly it is pwned by Oracle or MS SQL Server or DB2. It's about 1/3 the speed of the commercial systems, and the more cores you add the worse it gets. If they can keep Solaris working better than Linux, and gain market share, they have little incentive to give it away.
We switched from Redhat to Solaris 10 about 9 months ago and haven't looked back. Rock solid and cheaper than Redhat once you include support.
Sorry, but had completely the opposite experience. Spent three years switching our SPARC-Slowaris to RHEL on Itanium and x64 and it made life much easier. Red Hat support is usually excellent, when we have cause to use it, and we know we're using it a lot less than we ever did Sun's. We use a proper call logging helpdesk system that means we can go back five-plus years and giggle at all the rediculous problems we had with Sun. Our only annoyance is that Red Hat are not making RHEL 6 for Itanium, but we're working on migrating to hp-ux and/or RHEL x64 for those cases - Slowaris 10 on anything just didn't factor in our considerations. If you had problems with RH then I suggest you simply need to hire some people with the right skills.
"spent three years switching our SPARC-Slowaris to RHEL on Itanium and x64"
Don't worry about it, Matt. Someday you'll be able to look back and laugh.
Sun's big mistake was wasting about six years on a pissing contest with the other unix vendors.
It was Unix International (Sun, AT&T, and tiny Arix) versus Open Software Foundation (Digital, HP, Compaq, IBM...). They were too blinded by their own egos and Ayn Rand "free market" fantasies to realize that IBM had handed MSFT a monopoly that would destroy them all. That was the six years it took MSFT to get Windows and Office working well enough for the enterprise to tolerate it.
RE - Don Mitchell
"They desperately want an operating system with the stability of MS Server but without the complex and expensive licensing."
That, hands down, has got to be the funniest thing I have read on this site.....
RIP Sun well miss you
except for SPARC which was a mercy killing. Good god what a slow crap architecture after the late 90s or so. Burn white hot in hell SPARC (taking an eternity to calculate to the square root of 9).
Er, you obviously haven't heard how much Oracle are investing in SPARC, then...?
Er squared ...
... something that sounds very much like the old Sun, which prided itself in having the "2nd-largest microprocessor designer team in the industry", quote from (get the date of the quote from the document):
So if Oracle is investing in SPARC, they're as stubborn as Sun used to be, never stop throwing money at microprocessors. Looks like the two have really integrated now, worst of both worlds !
Opensource was not Suns problem
Sun never recovered from the dotcom burst of 2001 and instead of trying to remain competitive against IBM, HP, Compaq, and others, they were churning out all the same stuff as before. They then cut their support staff and the problems with their hardware started to take its toll. It was Fujitsu that really did them in however. They beat Sun at its own game. They made a more reliable Sparc machine with better after sale support. Sure Fujitsu servers cost more, but customers were willing to pay the difference once they realized just what it was that set the two companies apart.
Scott is rewriting history
Scott McNealy wants us to believe that dealing with ATT to merge SunOS with System V made
Sun a closed source company because, as a result, it could not open source Solaris.
Facts prove that at this time Sun was fiercely closed source, turning its back on its BSD background. The reality is that Open Source was considered an option only on the light of
the success of linux, that is after 2000.
But at the end of the eighties, Sun was closed source and tried to be too greedy.
Around 1988, Sun decided to unbundle the C compiler that was until then given with SunOS.
As a result, most people switched to gcc and many people who could live with free proprietary tools got a taste of Open Source tools (even if the term was not yet coined).
The reality is that the infusion of cash from ATT made Sun arrogant. That turned all the Unix industry against Sun. At that time, Sun was trying to push NeWS, its Network extensible Windows System.
The rest of the Unix industry pushed the Open Source X window System to kill NeWS and succeeded. That is why today most Unix System (except for Apple) run this X crap.
NeWS was based on the superior PostScript imaging model but free of Adobe proprietary code.
Had Sun opened the NeWS code, the story may have been different.
- iSPY: Apple Stores switch on iBeacon phone sniff spy system
- It's true, the START MENU is coming BACK to Windows 8, hiss sources
- Chinese gamer plays on while BMW burns to the ground
- Pic NASA Mars tank Curiosity rolls on old WET PATCH, sighs, sniffs for life signs
- How UK air traffic control system was caught asleep on the job