Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Sun Microsystems

No Solaris 9 for x86 272

Jon writes: "Unsurprisingly, LinuxWorld is reporting that Sun is not going to support Solaris 9 on PCs. The article cites a marketing suit who claims that the prevailing economic conditions account for this."
This discussion has been archived. No new comments can be posted.

No Solaris 9 for x86

Comments Filter:
  • by Jon Chatow ( 25684 ) <slashdot@jdforrester.org> on Wednesday January 09, 2002 @08:47AM (#2808816) Homepage
    Found here [nwfusion.com]. But is this good, encouraging the curious to move to free OSes when exploring beyond Windows, or bad, removing a great way of finding out about an OS that is easier to convince your boss to have installed?
    • > But is this good, encouraging the curious to
      > move to free OSes when exploring beyond
      > Windows, or bad, removing a great way of
      > finding out about an OS that is easier to
      > convince your boss to have installed?

      I think that not only is it good, but Sun *knows* it's good. One of the problems facing them at the moment is that companies whose products traditionally ran on UNIX (e.g. high end CAD) are producing WinNT versions too, because lots of companies are blinkered into the "everything on Windows" mentality.

      One way out of that is if people coming into (for example) engineering industries know that there is life outside Windows. SO what if people switch towards Linux in an academic environment, where Sun sell (sold?) a lot of stuff at heavily discounted rates? If it teaches them that 'doze isn't always the right tool for the job, then there may well be revenue coming their way when today's undergraduates become tomorrow's IT decision-makers.
    • It's not good. When starting to work with Solaris in my company I really enjoyed it to have a free Solaris8/x86 to install it at one of my PCs at home in parallel so I could hack it a bit and get more used to it by playing around with configuration options that I'd never dared to play around with on the systems at work.

      It would be _so_ good if one could also do this with Solaris 9 at home, provided your employer started to use 9 at work. At least Solaris 8/x86 is still there.

      Too bad this really fits with the news from today that Sun has removed the download links to Solaris 8. :-(((

      Because Linux at home on your Average Cheap Hardware doesn't help you to get used to SunOS. IMHO it was quite a clever idea from Sun to support Solaris on cheap x86 hardware and give it away for free, so more people had a look at it. And for you at home, it is always a good chance to know how as many as possible different systems look and behave. Yes, it's Unix. But if you've never seen Solaris/SunOS before and only hacked with Linux, you'd be amazed how different the system is.

      • >Sun has removed the download links to Solaris 8

        Although they removed the links to the download page, it appears that you can still download x86 solaris 8 from sun by just changing the 'sparc' to 'intel' on the download link [sun.com]

        Good thing too, I had decided to cobble a machine together to install solaris over the holiday break and had downloaded the HCL to make sure I was using stuff that was supported. I have the machine assembled, but I hadn't downloaded the CD images yet. Guess I'll be doing that tonight.
  • by jdh28 ( 19903 ) <jdh28@bigf[ ].com ['oot' in gap]> on Wednesday January 09, 2002 @08:49AM (#2808819) Homepage

    As this article [theregister.co.uk] on The Register points out, there are now no proprietary unices being actively developed on x86.

    Linux and the BSDs remain the only options.


    • An alternative thought to this is that Windows 2000, being a much more mature OS than NT, has taken over an adequate market share that unix has been pushed out of that niche. I've noticed most companies still tend to use unix for their back offices when they need massive amounts of processing power.. something that intel-based boxes just can't offer on a substantial level.

      I have to wonder, in percentages, how many companies would rather deploy, say, Solaris AutoClients as opposed to Cirtix & Thin Clients.

      Just a note: Solaris 9 for SPARC machines is going to go Non-developer beta next week.
    • by jsse ( 254124 ) on Wednesday January 09, 2002 @09:16AM (#2808900) Homepage Journal
      Time to celebrate that opensource has finally conquered x86. Where are the free beers we always talk about?
    • by zak ( 19849 ) on Wednesday January 09, 2002 @09:35AM (#2808961)
      UnixWare (now OpenUnix) is still in very active development. Check out the Caldera site.
      It's only the best environment to run Linux apps on a multiprocessor, so I see why The Register would ignore it :)
      • You may want to look deeper - deeper then press releases and PR pages..

        Go ahead to any Unix shop and ask them to write/port for you any application to Unix on X86 - sure, they'll be happy to do it for either Linux (most of them - RedHat), FreeBSD and Sco-Unix (or whatever Caldera calls it now)...

        Now - you might want to take a look at the price purposal - Linux will be the cheapest one, 2nd will be FreeBSD and ScoUnix - the most expensive one - I know because a company I know asked me to look a bit into it and asked around and I got several prices - the scene was the same on all of them.

        To make a long story short - it's dead, Jim - it might take a bit of time for SCO/Caldera Unix to die - but it's written on the wall - just like what Coherent Unix was on PC back then..
    • The Register points out, there are now no proprietary unices being actively developed on x86.

      Isn't Sequent Dynix/ptx still being developed?

      It's x86 Jim, but not as we know it.
    • by mpe ( 36238 ) on Wednesday January 09, 2002 @09:57AM (#2809028)
      As this article [theregister.co.uk] on The Register points out, there are now no proprietary unices being actively developed on x86.
      Linux and the BSDs remain the only options.

      I don't see how this is a bad thing. One major problem with proprietary unix systems is that you end up different proprietary addons.Thus incompatabilities between systems.
      Also remember that Sun's main business is selling hardware rather than software...
      • You're dead on. Sun's primarily business is hardware. So making an x86 port of solaris seems silly when they could spend the money/manpower of improving what is their best chance in the longrun -- staroffice -- of breaking microsoft's deathgrip. In fact, I'm a bit surprised they even want to make Solaris. Sun has the support capabilities to roadmap an end of life for Solaris and plan to release linux instead, and they could spend their time tuning linux for sparc processors. Solaris already has a lot of POSIX compliance (like its own pthreads library), and even sun sysadmins would take to linux -- I'd say Solaris and Linux feel like closer cousins from an administrators point of view than Linux and BSD.
        • I'm a bit surprised they even want to make Solaris. Sun has the support capabilities to roadmap and end of life for Solaris and plan to release Linux instead....
          This is an insane thought. Linux is currently (and will be for quite a few years yet) a big lose compared to Solaris in many areas. The SMP is crap in comparison, it has poor hardware support on Sun hardware (or even PC hardware, for that matter), poor I18N support, the list goes on. The Linux kernel is technologically many years behind the SunOS 5 kernel.
          • I can't directly argue that, since I'm very unfamiliar with the methodology behind either linux OR sun SMP features. I can say, however, that almost all applications I've used on both platforms when utilizing them as servers have inevitably run more smoothly and with better price/performance under linux. While I've used multi-processor suns, I have yet to utilize a multiprocessor linux box, so it is hard to make a comparison there, although I've always read heartening things about the linux SMP design that made me think it was well thought out and should be effective.

            I was aware that Solaris' IP stack, which was supposedly a derivative of the powerful/robust Mentat Portable Streams, was superior to Linux's, at least until the 2.4 kernel, but that should have changed in 2.4.X.

            Anyhow, poor hardware support? Obviously if sun spent their time improving linux for SPARC instead of building Solaris, that would change immediately.

            You'd have to cite more specifics otherwise to convince me (or probably most people who've used solaris and linux extensively, and prefer the latter). I've deployed hundreds of sun servers (mostly as firewalls, but many as mail/dns/web/etc servers) in my short career, and was an intern at Sun one summer as well. I've been much happier with my linux servers, even before the 2.4.x enhancements. What, more specifically, is so deficient in linux?
    • So if you want a Unix on x86, you have a choice between the free BSDs or Linux. Slashdotters rejoice. We just hate to see any good, cheap and well-supported OS bite the dust.
      The truth is that Solaris x86 was always something of a joke. Sun's primary business is selling SPARC-based hardware, and their priorities tend to favor that goal. Solaris x86 was the bastard stepchild -- half-arsed support, not very robust, limited app base. At one time (before Sun realized that they were losing ground to Linux) it was even hard to find the salesperson who was authorized to let you have a copy. Officially, the policy was "Solaris x86 is your upgrade path to Sun hardware!" But in reality the mindset was (and in some circles at Sun, still is) "x86 technology sucks. Junk it and buy our beautiful hardware!"

      That cheering sound you hear from the direction of Cupertino is that Javasoft people, who no longer have to pretend that Solaris x86 support is more important to them than Linux.

    • What about BSDi [wrs.com] (or BSD/OS, or whatever they call it now)? Yes, it's "a BSD", but it's also proprietary in a sense, no?
    • Data General has unix and they sell X86 base machines. These are not PCs, they are proprietary hardware with Intel processors.
      Does that count, or do they really mean "PC Architecture X86 Machines"?
      • Data General has unix and they sell X86 base machines.

        Data General do not exist, so they can neither have UNIX nor sell x86-based machines. They were bought by EMC; if you go to the old DG Web site [dg.com], you get taken to a site whose only mention of DG products is a link to the EMC Powerlink site, which appears to require you to have an account.

        The main EMC site [emc.com] doesn't seem to feature the AViiON systems; perhaps you can still get AViiON machines running DG/UX, but it doesn't look particularly easy to do so.

    • The only reason they made Solaris for x86 is that the federal government needed a fully POSIX compliant OS that ran CDE on cheap x86 hardware.

      Now that sun in making sub-$2000 SPARC workstations, there is no need for solaris x86.
    • Who would benefit from them?

      If there were no unix for a platform, there would be opportunities for a software vendor. There would also be opportunities for a vendor with a notably superior solution. There is *no* incentive for a hardware manufacturer to have tis own unix. One of the more important things linux has done has been to provide a common reference point--prior to this, it wasn't feasible for vendors to settle on a competitor's *nix as a standard., due to the admission involved. Now that there's a non-competitor that *is* the standard, it's economically more efficient to the hardware vendor to back that.

      hawk, who has a paper on this on his web site.

  • by ergo98 ( 9391 ) on Wednesday January 09, 2002 @08:49AM (#2808820) Homepage Journal

    The market conditions are that Solaris on Intel machines is a total failure. As another poster in another argument mentioned: The only people who Solaris on Intel machines seem to be just taking it for a test run, and then they go back to their real OS (be it Windows, Linux, FreeBSD, etc.).

    Just thought that was a little more honest than claiming it's the recession or Sept. 11th fallout.

    • We used Solaris x86 extensively at my old work (a university) as we couldn't afford Sun stations and an Intel solution gave us the flexibility of NT/Unix. Keeping with Solaris kept some continuity and the experience of Solaris on SPARC.

      Dunno what they'll do now, although I guess linux is certainly one option for them.

    • by Marcus Brody ( 320463 ) on Wednesday January 09, 2002 @09:11AM (#2808882) Homepage
      Yes, in terms of actual revenue Solaris on Intel is a complete failure. In fact, it is farcical. But I dont think Sun ever considered it to be a primary business venture - more of a 'loss leader'.

      And it has done this admirably. I learnt Solaris largely by playing around on my x86. It was fun - I really learned it like I wouldnt have done with a production system - man, I mangled that f*cker no end. Not that I could really do anything hugely useful with Slowaris that I couldnt do better with BSD/linux, but that wasnt the point. I have taken my experience with x86 Solaris onto using a 4500 workstation, where it is a good option for what we are doing. Who knows, if I hadnt had that first hand experience with Solaris, Sun may have been a few hundred thousand worse off.

      On the other hand, I doubt the experiment as a 'tester' was really worth the expenditure. The growing diversty in the x86 world was prolly the big killer, what with all these various chips and chipsets etc.
      • Yes, in terms of actual revenue Solaris on Intel is a complete failure. In fact, it is farcical. But I dont think Sun ever considered it to be a primary business venture - more of a 'loss leader'.

        Indeed, I've read many complaints arguing that Sun does a horrible job of optimizing on the Intel platform (hence "Slowaris") intentionally to make their own hardware look that much better. However that's a catch-22: It makes their hardware look better, but it makes their software look worse.

        • My brief experience with Solaris x86:

          A few years ago I blew away Mandrake something and installed Solaris 7 x86 on an old machines I had around (that was on Sun's HCL).

          The CDE install was certainly more responsive than the KDE v1 Mandrake had. The screen display was much crisper, and the graphics redrew faster with the Sun X server. And best of all -- Netscape was actually reasonably stable -- well, comparable to the Windows version at least.

          It was a little unusual of a box (SCSI, Matrox Millenium, supported by Sun), but at the time Solaris was a much more reponsive and "better" desktop OS than Linux. Maybe "Slowaris" couldn't fork Apache processes as fast, but you won't know it as an end user.

          The big problem with the product is that it caught no end of FUD -- from both the Linux crowd (who hated the [better] competition? and primarily have IDE and unsupported graphics like Nvidia) and from the Sun crowd (Sparc bigots, all of them, or well most of them).
    • I did some Oracle development on Solaris on (Sun-provided) Intel hardware back in '98/9. We would have been better to just buy a third party PC and install Solaris, but the people who bought the kit were suits and I was paid handsomely by the hour. :)

      First thing we did was start installing GNU stuff. Standard free tools like that, and keen competition in mid-range from Microsoft's excellent marketing teams, killed the proprietary UNIXes on Intel. Looks like the UNIX world is settling around BSD (and/or Mach) and Linux. Not before time.

    • I absolutely agree with you. Not only are people just taking a peek at it, and then going back to another OS - but they're also giving up due to lack of hardware support.

      The "out of box" experience with Solaris for x86 was pretty poor, IMHO. One of our former employees paid the $25 or whatever for a copy of Solaris for x86 when they had that promotional deal going - and we couldn't get it to support A) our Crystal sound chipset built onto our Dell motherboards, B) our nVidia graphics chipset, or C) our 56K internal modem cards. Of course, that wasn't even beginning to worry about such extras as USB support.

      Oh, I'm sure some Solaris fan will come along and tell me "You just needed to download driver X and Y from web site Z!" -- but that's not the point.

      I'd expect a commercial Unix to support basic devices like my video card right out of the box. I had much better hardware support in Linux, and I'm not paying anything for the rights to install it.
      • I've never had a problem with Solaris x86 supporting the hardware I had. However, my intention was not to use it as a workstation. I might also not that I've used fairly standard higher priced intel boxen. (Proliant stuff's mostly) Granted some people do use Solaris as a workstation OS, but I felt it strengths were in its stability as a server.
    • Not true. I run Solaris 8 x/86, and it's a beautiful, wonderful thing. I ran it as a server for over a year on my home DSL line. In that year on pacbells' network, I saw thousands of attempts to compromise the box, via scripts for windows and linux alike.

      Its unpopularity is beautiful -- no one scripts for Solaris 8 intel. I'm going to miss it a lot.
  • by y0bhgu0d ( 168149 )
    well that stinks... guess i'll have to buy a sunstation now.
  • Why can't they just support the OS with only the market-leading/most reliable X86 hardware? That would reduce their support costs and still enable people to put together a relatively cheap Solaris box. Or do they already do this?
    • Re:Why dont they ... (Score:5, Interesting)

      by Stephen Samuel ( 106962 ) <samuel@bc g r e en.com> on Wednesday January 09, 2002 @09:30AM (#2808947) Homepage Journal
      This was part of the problem with solaris-86 to begin with. I tried to run it on my laptop a couple of years ago, but found that there was no support for my (3com!) ethernet pcmcia card. Now, if I'm going to spend money on hardware just so I can run Solaris, I might as well just go out and get a used Sparc box. At least that way, I get some real support from Sun -- and I can use the same binaries as I use at work.

      In some ways, it's a bit of a chicken and egg problem. You won't get more drivers without more people using the OS -- but it's not worth spending thousands of dollars to create a driver that dozens of people are going to use... on the other hand, people aren't going to use the OS unless you have the drivers. . . . .
      rinse and repeat as necessary.

      Limiting the hardware you support even more than already would make the lack of users problem even more acute -- and the crowd (large handful?) of people using current hardware that would be orphaned by such a move would be up in arms about it. Far better to take your hit and essentially walk away from the X-86 market. Give end of life support to people running solaris 8 on X-86, and wean everybody else either onto real sun boxes (the preferred for Sun), or onto Linux -- which at least keeps them in the UN*X market.

      The other issue (as someone else pointed ou) is that Sun's primary interest in Solaris-86 was probably to keep people intersted in Unix-type operating systems, even if they only had commodity Intel boxes -- but Linux now does that so well, that it's easier (and cheaper) to put together Linux -> Solaris migration tools (done!) and Let Linux and the BSDs handle the X-86 market which they serve so well, already.

    • Re:Why dont they ... (Score:4, Informative)

      by anothy ( 83176 ) on Wednesday January 09, 2002 @12:16PM (#2809776) Homepage
      first off, remember that Sun's primary source of income is from their (very nice) SPARC hardware, not from Solaris (which you can often get free). they have no real incentive to work on an x86 version at all, unless it seems to be significantly helping their Solaris markent (and thus encouraging more SPARC hardware sales).
      note also that while your suggestion would reduce their support costs, it would not be trivial, and would likely not reduce them by nearly as much as you'd think. there'd need to be a certification process, and some detailed tracking of what cards of various types are/arn't supported, beyond just the base system. remember that when you by a "Dell Whatever" pre-built system, you have no real idea what exact video, network, or whatever card's in it; Dell (and all the others) think it's fine to change revisions of cards.
      • With the consumer Dell systems, you're right: the hardware changes from month to month. (It's not just in versions, either; you might get a completely different video or sound card in a system you order a month later.)

        However, Dell (and most other major manufacturers) also sells a line of machines that are guaranteed to have the same hardware for long periods of time. The machines are noticably more expensive and less poweful. But if you know you're going to be buying two thousand of them over the next three years, and you have to do your own OS configuration and certification for every new type of machine, it makes sense to pay more for less capability.
  • This hurts folks who want to learn on "cheap" h/w, but you can get a Sun Blade 64-bit workstation for $999 that runs the SPARC version of Solaris, so there are options for developers and those who want to "learn" Solaris and e-Bay is full of old SPARCs that are *very* indexpensive.

    Solaris x86 was a dog on uniprocessor systems and multi-processor boxes aren't worth the cost when you can get a decent SPARC *blade* system for $999 and have 64-bit processing power.

    IA-64 is still far off, and you can bet that Sun will be there when that technology is actually released and more mature since they *have* to compete with M$, IBM and HPaQ on enterprise turf where dumb suits and admins think of "plug" when they hear "spark".

    As a Solaris daily user, I'd rather run Linux or QNX on PC h/w than Solaris anyway. Better updates to match h/w advances along with solid performance on single-chip boxes.
  • Uhm.... why do they say "chips from Intel, Corp."?
    What about Athlon/XP/MP or Hammer?
    K, Solaris 7 ran bad on my Athlon 500, but at least it ran there

  • Heh (Score:2, Funny)

    by mESSDan ( 302670 )
    Sun gave developers their first peak...

    Now, is this just a typing error, or is Sun *that* good? ;)

    Me, I must've peaked early.

  • sun used to say "we do solaris and only solaris "

    they where proud of it as all the Unix vendors where selling NT

    now they have linux and solaris that makes 2 in my book
    (granted they are both unixy)

    I wonder what the SUN sales Spin is going to be now


    john jones
    • The Sun spin remains just about the same. Instead of saying we just do solaris they say they can sell a complete line of binary compatible machines ($1000-$10,000,000) running solaris. Then they so, "Oh, and we are also a leading vendor of Linux machines [in the niche cobalt is in]". So see, we are also good guys who like Linux and all that, but for real beefy servers you want solaris. That's what their spin is.
  • Sun doesn't make their money on the OS, they used to but not anymore. They make it on hardware and on support both of which are pricey but worth it in my opinion. Solarisx86 has always been the redheaded stepchild of Sun anyway.
  • Yeah, "prevailing market conditions". That, and it just plain sucks. Not only does it lack the kind of enterprise-class support for SMP that it's USparcIII cousin has, but it has absolutely pitiful hardware support. Hell, it even makes BeOS look good.

    I'd say that even Linux would be a better choice for your x86 machine than Solaris anyway.

    • mmm, no. having run Solaris/x86 and BeOS for years, Solaris, while quite poorly supported on x86 hardware, still soundly trounced BeOS. which is not to say it was very good: all the BSDs and Linux beat Solaris at least as much as Solaris beat BeOS, but be fair.
      also, i'm curious what sort of SMP problems you had. i ran it on quad-processor boxes, and it performed quite nicely; quick and stable. the biggest problem in my mind was always the application suport, which was almost non-existant.
  • A person in the article states, "The rising popularity of Linux in these communities is now doing this job for Sun and reduced the burden of promoting Solaris."

    When it comes to the x86 platform, Linux is ubiquitous, and there are thousands of precompiled binaries available for it. AFAIK, unless one is willing to compile everything from source, the number of apps available for x86 Solaris is much smaller.

    This is nothing more than the free market at work. Consumers choose the best product for a job based on ease of use, availability, and other factors. For most x86 users, there is not enough of a difference between running Linux and Solaris to justify the support of the latter.

  • by beezly ( 197427 ) <beezly.beezly@org@uk> on Wednesday January 09, 2002 @09:16AM (#2808901) Homepage
    Another interesting point to note is that Sun have not said that Solaris x86 is being canned. A t the moment they have just deferred the release of an initial FCS version of Solaris 9.

    One of my colleagues suggested that perhaps Sun are testing the market, to see how people respond to a threat against future releases of S9/x86. If they wanted to get rid of Sol9/x86 then surely they could just come out and say so, but they haven't done that. Perhaps there is more to this than it initially seems.

  • by upstart1234 ( 201823 ) on Wednesday January 09, 2002 @09:29AM (#2808944) Homepage Journal
    I keep seeing people posting that if you really want to run solaris 9 that you should just buy a sunblade 100 for $995.. sure thats the base unit cost but just to add a network card on suns site you add $600
    YES FOR A NETWORK CARD.. that network card better be one designed by god for that price... sun hardware is way to costly for a student that just wants to learn to use it.. not every school has sun boxes laying around for use.
    • by Quaryon ( 93318 ) on Wednesday January 09, 2002 @09:58AM (#2809031)
      Huh? The networking is built in, as pointed out by another reply. Also, the big advantage of the Sun Blade 100 systems is that you don't need to buy any other Sun hardware - they take commodity PC133 ECC SDRAM DIMMS, standard IDE hard disks, standard PC monitors and use a USB keyboard and mouse.. so don't look at Suns' inflated prices for these components.

      It cost me around £1200 for a fully working 64-bit system with 2Gb RAM at home (the boxes are much more expensive here in the UK as usual) which is easily comparable to a "reasonable" development-standard PC workstation with the same levels of stability.

      (I have two - one at work and one at home - they're great - try them!)

    • I'm guessing that price is for a single NIC.

      ** EACH AND EVERY ** Sun system has always had networking built-in!

      "The network is the computer"... ring a bell? they mean it.
      • ** EACH AND EVERY ** Sun system has always had networking built-in!

        Just to amplify this, every Sun box I have seen back at least to the IPX (and probably 3/50s as well) have had networking. The MAC address is on the NVRAM (or else read from a thing that looks like a transistor) on the system board.

    • If you are really a student, Sun offers incredible discounts on Hardware. Also comparing that Sun Blade to the O2 I bought 3 years ago for $6,000, sounds like a heck of a deal to me.

      Besides, how much should a computer cost that you can use for more than ten years?? I have a SparcStation 20 here that is older than most of the Slashdot members. It went from App server to print server, to development workstation. If Sun hadn't bundled that stupid Hot Java browser, I wouldn't even notice that this machine is kinda slow.

      If you run Solaris on x86, you can plan on paying $500 three or four times before the Blade gives out. Goes to show that some people know the cost of everything and the value of nothing.

      If OS'es were houses, Microsoft would be a trailer park.
  • After installing Solaris 8 on a leftover box last year I can tell you two things:

    1. There's no substitute for 64 bit arch
    2. Only the most common X86 hardware is supported

    First off, to be fair, the box had a K6-II@400 mghz so I didn't expect it to blaze in the first place. But I'm used to the performance of the E10k's we have layin' here at work as well as various 6500s and 4500s so I was a bit disappointed. It was cool for a while though. The box is back to running whatever linux distro I feel like messing with. (I should put a crontab entry to fdisk every Saturday :P).

    I had to go through the parts box for a video card that was old enough for Solaris to like (I don't remember Trident or Virge something). No Voood Doo or Rage goin' on here although I suspect some patch might work later after the install.

    Sun did a nice thing releasing Solaris 8 for x86. I certainly helped me become more familiar with that OS as a whole. I wouldn't recommend, however, using the x86 version on a production intel machine. There are better OS's for 32 bits :).
    I don't mean to be putting down Sun's efforts. The gave us Solaris 8 (for free even). I just don't think we're going to miss much without Solaris 9 x86.

  • by BenjyD ( 316700 )
    I used Solaris 8 on intel for a few months and I have to say it was pretty nasty. Very little hardware support, poor performance and huge difficulties getting software to work, as porting to solaris is not exactly a high priority for developers.
  • Delay ? (Score:4, Interesting)

    by sconest ( 188729 ) on Wednesday January 09, 2002 @09:57AM (#2809030) Homepage
    I thought they were delaying [cnet.com] it (with no future date announced).
    • They were -- the Register was simply associating delay with death, going by the rapid development cycle of Intel-based hardware.

      By the time they get around to releasing, there will be new chipsets, new peripherals, etc. all not supported.

  • by Waldmeister ( 14499 ) on Wednesday January 09, 2002 @10:04AM (#2809057)
    A Sun engineer told me yesterday, that Solaris 9 for x86 will be deferred some time, but _not_ eol'ed.

    There is currently a beta for x86 and a release is still planned and worked on.

    I believe this engineer quite trusworthy, especially more than a Linux gazette...

    Another interesting piece of information from this source: they are stopping the possibility to download Solaris 8 x86 from their webserver, but you have to buy the media kit.
  • by 0xA ( 71424 ) on Wednesday January 09, 2002 @10:10AM (#2809077)
    Solaris 2.6 x86 had a pretty defined market. If you had a bunch of Sun infrastructure type machines and needed workstations for admins and developers you didn't have to go drop 30k. You just grab a 2500 dollar Compaq and fire it up.

    Of course half the software you needed didn't run on x86 and hardware support was abysmal (couldn't get v8 to talk to my 3C905, I mean c'mon here). But damn that was a lot of money you just saved.

    Then Sun decided to release their Ultra 5 workstations at 6k a piece or so, IIRC. The market for Solaris x86 went **POOF** in about 4 seconds. The damn things are real live UltraSparcs and they work like a hot damn.

    Sun made the usual moves to try and spark interest, gave it away free, devoted new marketing resources to it etc. But it didn't catch on, unless you really needed Solaris on your x86 for some reason most of us tried it for 2 days and ran right back to linux or *BSD as fat as we could.

    I mean really, with a nicely setup Blade 100 going for $2,450 at store.sun.com [sun.com] who would ever bother with a half suported stepchild?

    • I've got a sunblade 100 - unless you need the fancy video card - which I don't since I just use mine for coding, running weblogic (dev), and oracle (dev) - the $1000 model will work just fine.

      It uses PC133 ECC SDRAM, which does not cost a lot of money (I paid $63/512M stick in November). 3x512M + 1x128M, ah, sweet necture of the gods....

      Also, think about adding a SCSI controller and HDD if it is for something other than development. The IDE drives won't cut it in a multi-user environment. Should set you back about $300 for an Adaptec 160 controller, and about the same for a SCSI-160 drive. The IDE drive I got was only 15G, not sure what RPM....
  • Overdue Decision (Score:4, Insightful)

    by 4of12 ( 97621 ) on Wednesday January 09, 2002 @10:18AM (#2809108) Homepage Journal

    From a business perspective, I think this makes a lot of sense for Sun.

    A few years back a friend tried to create a "UNIX laptop" for the purpose of having a portable roadshow platoform for a scientific code we have that was developed primarily on Solaris 2.5 and SPARC. At that time he found that Solaris/x86 was a lot of hassle to deal with and that Linux 1.2 was a better solution for him.

    I think the resources spent on Solaris/x86 would have been better invested in bringing out the UltraSPARC III sooner and in further expanding utility of their big servers.

    Am I missing something obvious in the following observation about the market landscape?

    • x86 is further dominating the desktop, even now in the UNIX circles, where Linux/x86 offers price/performance ratios that *NIX/RISC cannot match
    • big 64-way 128-way machines with high throughput are safely owned by *NIX/RISC, as IA64 development has been a fiasco (Intel would do better if they just swallowed their pride and brought out the Alpha 21364 under the house brand).

    From my perspective, Sun would do well to find as many ways as possible to make Sun servers attractive in LANs of Linux/x86 desktops. The arena of high capacity servers is where x86 falls short and Sun shines. Make the most of it.

  • It's a complete fscking joke! God forbid you want to install a decent raid controller (say 64bit 66MHz for example), even if you find one that is "supported" by Sun and the vendor, I bet money you'll be on the phone with the vendor for three days trying to find a working set of drivers. It's dead slow on Intel. Install DOS, it's faster and has better support.

    I get stuck dealing with it because the poor fools we support absolutely MUST have "development" boxes that mirror the production boxes. All the production (Oracle 8i DB) is Solaris on Sun hardware. We can't afford Sun boxes for testing, but the platform has to be the same, hence the need for Solaris on Intel: cheap Sun development boxes. Putting Oracle 8i and Solaris 2.7/2.8 on Intel is like trying to install OS/2 Warp on a Commodore 64. Managers get pissed when you bill them a shitload of time (2+ days on some occasions) just for a working OS install. Especially when you write "Use Linux next time" in the comments field. ;)

    Every chance I get I hammer management relentlessly, without pity or mercy, about what a shitfest this OS is on Intel. There is NO excuse, Linux can be made to simulate a Sun environment with precious little effort. Thank GOD that Sun finally decided to can this thing. Now I get to sit back and laugh hysterically since they have no choice except to use Linux. SuSE + Oracle = 10,000x faster performance on Intel than Solaris.

    In case you couldn't tell, I have enough frustration energy from dealing with this OS to light up a small star system for a few years...

    I've nothing against Solaris on SUN hardware, mind you. It kicks a hell of a lot of ass there. There's something very nice about an OS optimized specifically for the hardware it runs on. Must be why Mac users are always smiling (or it could be the drugs they are on, what do I know...)
    • Hmm... 2 days * 8 hours * $100/hour = $1600

      Looks to me like it'd be cheaper to buy a low end Sun workstation. But pool the groups and you could buy one larger Sun machine and install multiple instances of Oracle on it. That's the way our dev environment works, we have several dozen Oracle instances on a 4 proc HP N-class.

      I absolutely agree with your management that test systems should mimic production systems as close as possible. Linux is obviously not a close approximation, and it is even arguable if Solaris for x86 is close enough.

      As you point out the labor costs of trying to identify and fix an issue which is different between development and production will more than exceed the added cost of buying the hardware.
    • ...as it is not supported by Oracle 9, so you've probably had growing ammunition for your Linux switch for some time (but not SUSE - anything but that!).

      People say that the Sun X server is more stable than xfree86, although I haven't seen that to be the case on Solaris-x86. Still, it would be great if the xfree86 people would agree to supply the X server (esp. for an exchange of GPLed source code).

      In fact, Sun should seriously evaluate:

      • A Linux device-driver (-compatible) interface for the kernel (but this may not work with the gpl - but can anyone litigate if they include just the interface, but no drivers?)
      • Adopting one of the Linux boot managers (theirs is REALLY ugly)
      • Even GPLing enough of the kernel to possibly integrate SGI XFS, IBM JFS, or the others into the native Sun kernel (something with dynamic inode creation)

      I assume that all the source code has been publicly available for all this stuff, but no one could work on it because of the NDA...

      Sun should adopt more creative cost-cutting measures to keep it alive.

  • this would have been reported differently.

    For example, if some Linux distribution decides to stop supporting Alpha (well, I *do* work for Compaq!) or Sparc, the media would be sounding the death knell for those processors.

    So... where are the obits for x86? [smile]
    • Because the most common operating system for the x86 platform is not Unix. It's Windows. And Windows still has one of, if not _the_ largest market shares of any OS.

      If Microsoft suddenly announced that they were dropping support for x86, and porting WindowsXP and all future releases of Windows over to PowerPC, or SPARC, that _would_ be a death knell.
  • And the performance of it was TERRIBLE !

    The price of Sparc hardware, especially the AX engine stuff is soooo cheap now!! For under $1000 you can get ATX formfactor sparc processor computers.

    Whenever I have conversations with my associates and I hear them bragging/bitching about "lets write a new OS" -- my first argument against is "device driver hell"

    NO ONE has that much freetime in their lives to write driver dujour for hardware X

    Look at the limited set BEOS supported and ask yourself the same question
  • by Cylix ( 55374 ) on Wednesday January 09, 2002 @12:08PM (#2809722) Homepage Journal

    About a year ago I decided it was high time I get a little more experience with this demon known as Sun. At that point I was a hardcore Linux/X86 kinda guy with a love for Digital Unix as well.

    So I pulled a Proliant from the back of the NOC and began installing Solaris 7 X86. Note, these compaq systems are Solaris certified (ie, every piece of hardware we had will work). The install went flawlessly and the box was up on the network upon completion. Granted solaris has a few *extra* features in inetd, but anyone with some sense can chisel that down to what is needed.

    I could go into detail on everything I've done with the system, but the there really is only one bottom line. Solaris isn't a bad operating system at all. As long as you have all of the dependencies, most applications compile fine. (well, what I've used on the server end).

    Sun support for non-customers has been fairly well. They release patches and updates frequently (not sure if its too frequently, but at least they fix their problems).

    I've been happy with this operating system and I'm going to miss not installing and using 9.

    The system is not without faults and I'm not an expert. Like any other piece of software, there will be times when it will frustrate the hell out of you. Thus is the nature of technology and if I damn Sun for it, I have to damn everyone else. (oh hell I do that all the time)

  • I have heard from numerous people that Solaris x86 is slow and hard to set up. I use an x86 box to JumpStart my Sparcs and it is definitely not slow! Of course my machine is a dual Celeron rig with 768 MB of RAM and 2 UDMA 100 drives. I find that Solaris x86 performs extremely well given that you install it on hardware that fits the HCL. And that is where the problem lies with people installing Solaris x86, I read through posts on alt.solaris.x86 daily and see people trying to install Solaris with any hardware they just happen to have then bitch about it not working or being too slow! With any OS there is a learning curve and I guess some people just aren't up to the task. As far as the Blade 100 argument goes, we have 13 Blades at work and 4 of them had to have either system boards, CPU's or other components replaced. The performance of a Blade 100 sucks without a memory upgrade due to the excessive paging in the base configuration (128 MB). We dropped in a second 128 MB stick so that we could install Sun Management Center and the paging virtually stopped! So I wouldn't go around saying "buy a Blade 100", I won't! I think Sun's management is missing the point with Solaris x86 and the "bottom line". Yes it costs them money to produce it, but if you want to expose the maximum amount of people to it, what easier way than to make an x86 compatible version. Admittedly it might not support some hardware but at least you could use it for some things (like JumpStart servers) and use it as a tool to convince management that Solaris is the way to go. From a learning standpoint it is far easier to build an Intel box that will run Solaris than to buy a Sparc (remember most people learning Solaris do not know the "ins and outs" of Sun hardware). Hopefully Sun will "wake up" and continue to produce Solaris for Intel, even at a loss.
  • What a pity. The advanced installer Sun used to setup Solaris 8 on x86 will truely be missed. So modern, so flexible, so truely up to its task to adapt to the hardware and harddisk partition tables.

    *sniff*. Now no-one will ever experience the true joy of installing Solaris on x86 using this 22nd century technology...
  • I couldn't understand what made Solaris x86 such a great idea... when the HCL could be printed legibly on a postage stamp, it just seemed to me that running Solaris on your x86 machine made about as much sense as running Windows NT on your Sparc 5...

    Now, if they would have done some emulator work and given (slow) binary compatability, I could see why one might want to spend a kilobuck on a cheap PC instead of five on a cheap Sparc, but as it sat, I couldn't make heads or tails of what market it was aimed at.

    • I couldn't understand what made Solaris x86 such a great idea... when the HCL could be printed legibly on a postage stamp, it just seemed to me that running Solaris on your x86 machine made about as much sense as running Windows NT on your Sparc 5...
      For me, the point of having Solaris x86 was that admins could have solaris desktops using (quality) commidity PC hardware, and have the exact same user environment on the desktop PC as on the servers we administer.

      Hardware compatibility was actually pretty good, with 2.6 and 7 each supporting a good range of major-brand (Compaq, Dell, IBM, etc) PCs, and even a number of laptops.

      Solaris x86 wasn't something you would generally use to deploy a server (Sun did want to sell a few Sparcs), but it did serve real purposes.

      Now, if they would have done some emulator work and given (slow) binary compatability, I could see why one might want to spend a kilobuck on a cheap PC instead of five on a cheap Sparc...
      Given that Sun makes their money off the Sparc hardware, to the extent that they give away the operating system, that would be a bad move for them.
  • Now, I do like Solaris, but never found the x86 implementation to be worth the trouble. Since Solaris is mostly source compatible with Linux or *BSD anyway, the point is moot.

    If you want a real machine, buy a sparc, otherwise just keep using the free unices on x86...

The IQ of the group is the lowest IQ of a member of the group divided by the number of people in the group.