When Sun Microsystems' Founders and Former Employees Hold a Reunion (infoworld.com) 36
Last week Infoworld reported on a reunion of more than 1,000 former employees of Sun Microsystems including all four founders of the company -- Andreas Bechtolsheim, Vinod Khosla, Scott McNealy, and Bill Joy -- at just their second reunion since the 2010 Oracle acquisition.
Prior to the formal festivities, the company founders met with a small group of press persons. Pondering recent developments in computing, Bill Joy, who is now concentrating on climate change solutions, recalled that Sun tried to do natural language processing, but the hardware was not fast enough. Regarding the emergence of the iPhone, Joy said the advent of mobility and data networks has been transformational for society. He noted that Sun had that kind of vision with Java ME, with Sun trying to do programmable smartphones. "But the hardware was just really nascent at the time," Joy said. Machine learning, though, will be as transformational as the smartphone, he added.
McNealy emphasized Sun's willingness to share technology, such as the Network File System (NFS), which helped to bring about the open source software movement now prevalent today. "We didn't invent open source but we [made it] happen. We were the leader of that parade." Asked if Sun should have moved from Sparc Risc processors and Solaris Unix to Intel processors and Linux, McNealy said he did not want to talk about mistakes he had made as Sun CEO but such a switch was not what Sun should have done....
Among those proudest of Sun's achievements was Sun founder and CEO Scott McNealy, who, taking the stage, had some sharp words for Facebook, which now occupies one of Sun's former Silicon Valley campuses, without mentioning Facebook by name. "I remember some company moved into one of our old headquarters buildings," McNealy said. "And the CEO said, we're going to leave the [Sun Microsystems] logos up because we want everybody in our company to remember what can happen to you if you don't pay attention. This company could do well to do one-one-hundredth of what we did."
McNealy emphasized Sun's willingness to share technology, such as the Network File System (NFS), which helped to bring about the open source software movement now prevalent today. "We didn't invent open source but we [made it] happen. We were the leader of that parade." Asked if Sun should have moved from Sparc Risc processors and Solaris Unix to Intel processors and Linux, McNealy said he did not want to talk about mistakes he had made as Sun CEO but such a switch was not what Sun should have done....
Among those proudest of Sun's achievements was Sun founder and CEO Scott McNealy, who, taking the stage, had some sharp words for Facebook, which now occupies one of Sun's former Silicon Valley campuses, without mentioning Facebook by name. "I remember some company moved into one of our old headquarters buildings," McNealy said. "And the CEO said, we're going to leave the [Sun Microsystems] logos up because we want everybody in our company to remember what can happen to you if you don't pay attention. This company could do well to do one-one-hundredth of what we did."
More good than bad (Score:1)
Sun really did seem to be one of the last big tech multinationals that did more good than bad in the world.
I guess some might argue IBM holds that mantle now. But the score card for the industry does not look so great recently.
We're definitely past the visionary stage of this revolution.
Not an employee but a fan (Score:2)
Re: (Score:2)
I actually interviewed to be an employee mainly because their product was good UNIX stuff back in the 90's and maybe beyond.
I remember way back then there were three "cool" workstation companies to work for:
DEC, HP and Sun.
Fast forward to now, and DEC and Sun are dead. And HP is a totally different company now.
Ironically, IBM was never "cool" to work for, but they still sell POWER RISC processors and their own Unix variant, AIX.
Re: (Score:1)
Ironically, IBM was never "cool" to work for, but they still sell POWER RISC processors and their own Unix variant, AIX.
And run Redhat, SuSE, and Ubuntu Linux [ibm.com] on their mainframes.
Remember, kids, no one ever got fired for running Linux on IBM!
Re: (Score:2)
I'm pretty sure Apple is the only company still selling a "UNIX workstation" (in that MacOS is BSD-ish under the GUI). Maybe places like System76 count?
Re: (Score:2)
Share the sentiment, so this is basically a concurrence.
My links to Sun were even weaker than yours. Main one was borrowing some PERL from an ex-Sun employee (for one of my earliest websites (which is still up and running)). We were working together at an Internet startup before the first bubble burst. Sharp and a nice guy. RIP.
Also I used Sun workstations and Sun-created software over the years. Oracle's destruction of the software bothers me more than most aspects of Sun's departure...
Sun just wasn't a me
Not only NFS and Java.... (Score:5, Informative)
Re: (Score:1)
Moving to X86 exclusive would have been a mistake (Score:2)
If you look at it the SPARC platforms consistently sold better.
Their main issue was missteps in SPARC CPU design during the UltraSPARC timeframe and later with processors like Rock.
This meant their CPUs were consistently underperforming but even then platforms like Niagara did well enough.
In the end the main issue was what they offered was in several ways inferior in price/performance with Linux/X86 and their original main claim to the market was gone as UNIX became widely available. Much like SGI with NVID
Re: (Score:2)
Re:Moving to X86 exclusive would have been a mista (Score:4, Insightful)
Sun had some great technologies. But they had.... a lot of marketing and business model problems. That whole step into from BSD into AT&T based UNIX was aimed at selling the company to AT&T, and that never happened. Coupled with the switch from "SunOS" to "Solaris", the switch was an enormous mistake and turned a lot of engineers and developers away from Sun to Linux, which was much more BSD-like. When we discovered we could get free compilers by simply running Linux on Sparc hardware, well, there went the profit margins for Sun.
Do not get me *going* on how they screwed up Java releases and marketing and packaging. The simply fact that they couldn't keep the numbering straight was a hint at the conflicting "big picture" ideas that just destroyed them. Was it Java 4, Java 1.4? Java 8? Java 6? Or Java 1.6.0_u203? Stop inventing numbering schemes, no one pays you to do that!!!!!
I still remember the excitement about Solaris, "no, you can't run SunOS on the new hardware, you *must* switch to Solaris!!!" And then learning that Sun's own engineers insisted on sticking with SunOS, and Tatung taking advantage of vendor licensing to port SunOS to the new hardware. I laughed for days as Sun admitted it was possible, and finaly releast 4.1.4 that ran on both hardware platforms.
Re: (Score:2)
That whole step into from BSD into AT&T based UNIX was aimed at selling the company to AT&T, and that never happened.
When they abandoned BSD was when my company decided to abandon Sun.
We switched to FreeBSD on x86 instead.
Their move to SysV never made any sense to me. I never heard the AT&T acquisition story before, but that would explain it.
Re: (Score:2)
Re: free compilers (Score:2)
My boss was cheap, very cheap, and there I was with 670's, 690's, a bunch of Sparc10's and SparcStations.
The Sun OS included an assembler!
Solution:
1> Use the assembler to compile a mini C compiler.
2> Use FTP to download the GCC source.
3> Got to coffee, and lunch. It's going to be a while.
4> Profit!
Re: (Score:2)
If you look at it the SPARC platforms consistently sold better.
Yes, because their x86 machines were always crap. They didn't behave like their SPARC machines.
Their main issue was missteps in SPARC CPU design during the UltraSPARC timeframe and later with processors like Rock.
Rock would have used 250W, yowza! But it wasn't just Rock. Niagara sold like a turd, too. If your task didn't break up into many threads it was worthless, and the cost didn't justify the performance either. Sun failed to anticipate clusters. FAIL!
In the end the main issue was what they offered was in several ways inferior in price/performance with Linux/X86 and their original main claim to the market was gone as UNIX became widely available.
It was inferior at any price.
Re: (Score:2)
If you look at it the SPARC platforms consistently sold better.
Their main issue was missteps in SPARC CPU design during the UltraSPARC timeframe and later with processors like Rock.
This meant their CPUs were consistently underperforming but even then platforms like Niagara did well enough..
It's difficult to decouple this from the shift in the Intel x86 segment.
When UltraSPARC II shipped it made heavy use of optimizing compilers to make those multi-issue pipelines work, and the whole RISC vs. CISC game was still in play. The big design mistake in the UltraSPARC timeframe was failing to use ECC cache. They gave it parity bit error detection, and then used chips that required ECC to meet the published reliability spec. So it could detect bit flips, but do nothing about them. This led to the
Re: (Score:2)
The lack of error correction on cache in the initial F15 CPU boards meant that a cosmic ray burst could crash a machine or cause various types of memory corruption. This design choice was neither smart nor clever; within a few weeks of two F15 frames going live we had to replace all the non-ECC cache CPU boards.
Re: (Score:2)
This led to the infamous ecache parity error kernel panic in the bigger systems.
The lack of error correction on cache in the initial F15 CPU boards meant that a cosmic ray burst could crash a machine or cause
various types of memory corruption. This design choice was neither smart nor clever; within a few weeks of two F15 frames going live we had to replace all the non-ECC cache CPU boards.
It always amuses me when people invoke "cosmic rays" as their stray random radiation event of choice. It's kind of almost invoking the supernatural, to imagine that one 10^12 eV particle per m^2/sec arriving from the other side of the universe at just the right transistor. I mean yes, they do occur, and they can definitely cause bit flips. But... You really need to think closer to home. Consider that most of the lead on planet earth is the decay by-product of uranium. Hence, most lead has a tiny minus
Re: (Score:2)
I didn't choose "cosmic rays" as my "random radiation event of choice" for two reasons: first, I wrote "cosmic ray burst," the origin of many, many more particles than a single cosmic ray. A physics expert (if there are any in this crowd) may wish to comment on the typical number of particles emanating from a single cosmic ray impacting the atmosphere.
Second, the phrase "cosmic ray burst" came from
Re: (Score:2)
"Before the ascendancy of 32-bit x86 and later 64-bit x86-64 in the early 2000s, a variety of RISC processor families made up most TOP500 supercomputers, including RISC architectures such as SPARC, MIPS, PA-RISC, and Alpha." (https://en.wikipedia.org/wiki/TOP500 [wikipedia.org]) Sun didn't see the on-coming express train of the x86/x86-64 and it crushed them.
Re: (Score:2)
No, it wouldn't have been a challenger to Winders. MS had the manufacturers sown up and without that kind of scale, Sun could never have competed. I don't know that there was a way to save Sun given where the marketplace is now.
One think I will never forgive Sun for is selling out to Oracle. Uncle Larry is malign force in DB and anything else Oracle touches.
Re: (Score:2)
One think I will never forgive Sun for is selling out to Oracle.
They had no choice. It was either sell out or shut down.
Re:SunOS on the desktop! (Score:4, Interesting)
I always have a soft spot for the history-what-if of Apple and SGI merging sometime in the mid 90s. SGI gets a desktop market and a really good user interface, Apple gets serious UNIX know-how and access to superior graphics processing than was available in desktops.
I wonder how a Sun/Apple pairing would have worked along the same lines in the same era. Especially if it was something that was done before Microsoft released Windows 95. IMHO, Apple had a huge usability and UI quality advantage before Win95 was released and somehow it seems possible if you put a better vertical computing platform together with a superior interface, it prevents Microsoft from gaining as much momentum, especially on the server side.
*Maybe* it leads to a world with server/serious computing running on Apple/Sun equipment and Microsoft mostly an application company with a commodity desktop operating system, but not able to pull off serious business applications.
Great hardware, but expensive (Score:5, Interesting)
Pisser was I was a part time sysadmin. I had my own projects and deadlines, outside of being a sysadmin. Took 20 minutes or so to update each Sun to it's maximum RAM, and it was all my own time
Sun made great machines, but goddamn they sucked to work with.
/ that was the company where I could whine for many more than 3 slashies
Ooops, thought this was fark. My bad with the slashies
Re: (Score:2)
In general it was a bad idea to buy anything but the base machine from Sun except in the earliest days. Not long after they moved on from VME to SBus, you could get everything cheaper and better from someone else, even processors. The Ross HyperSparc beat the Sun processor offerings like small pieces of meat until the UltraSparc came out.
Re: (Score:1)
R.I.P. Sun Microsystems (Score:3)
I miss Sun. When I was getting into UNIX and Linux type operating systems, they seemed like the only looming tech giant that seemed like they were on 'our' side. I used to really believe in them, and it's sad that they've been eviscerated by a monster that now occasionally wears its skin.
Solaris 7 on x86 was my first UNIX, and even though I didn't have the best hardware for it (It was my first PC build also, and I didn't know anything about HCLs or the like) I still kept using it because I fell in love with it. My computer could dual boot Solaris and Windows95, and you can imagine what kind of a contrast that was.
Scot McNealy's quote is spot on, btw.
"the network is the computer" (Score:2)
...was Sun's slogan long before Microsoft discovered the Internet and later pushed the term "cloud" for Internet-services.
Sun's mistakes (Score:4, Insightful)
Sun created Java but never had a good plan on making money with it. In fact, without IBM and others helping it Java may have never even gone server side and wouldn't be used at all today.
They had some great Unix servers but saw Microsoft as their enemy when Linux was their true rival by basically giving people a Unix-like OS but without the expensive hardware and licenses.
They wanted to conquer the desktop but where never able to deliver a compelling UI in their products. It took them long enough to make Netbeans usable, they were so stuck in their old ugly Unix ways.
They wanted to go mobile, but Java ME was never successful. Again, another company came to the rescue (Google) and created Android. Sun was again left behind.
There was a lot to love about Sun, but they were essentially another Kodak, a company that saw its own death coming but was unable to transform fast enough to stop its own demise. In contrast, I'd say that's Microsoft's greatest strength.
Re: (Score:2)
Java did make some inroads into the embedded space back in the day.
Every Blu-Ray disk uses Java (for the menu system IIRC and also for some of the DRM). Some cable platforms use Java for various interactive things via the OCAP stuff. And there are probably more places Java ended up that I don't know about.
Re: (Score:1)
riiiite (Score:2)
"We didn't invent open source but we [made it] happen. We were the leader of that parade."
funny, rewriting history. if they actually were on-board with real OSS they might still have been around.
SunOS was the 1st port (Score:2)
Way back when SunOS was what you ran on Sparc, all the free software would compile & run on it. Any other unix was a 2nd consideration to the developers. Some things might target another platform (Sam Laffer's (?) FAX system for SGI) but there was almost always a SunOS version. GNU developed autoconfig and that standardized compiling software. No more hand editing Makefiles or shell scripts written in the author's exclusive style. At that time (1995), I was supporting SunOS, Solaris, HP-UX, Irix (SGI