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

 


Forgot your password?
Close
typodupeerror
×
Unix Operating Systems Software

The Future of Commerical Unices? 88

Posted by Cliff from the gaze-into-your-crystal-ball dept.
An anonymous reader asks: "I was recently wondering about the state of the commercial Unix world and what their plans are for the coming releases. I know Sun just released Solaris9 and HP is killing Tru64. But what about others like IRIX, HP-UX, SCO, etc? How has the rise of Linux affected these companies plans?"
This discussion has been archived. No new comments can be posted.

The Future of Commerical Unices? More

The Future of Commerical Unices?

Comments Filter:
  • IBM with AIX and Caldera with OpenUNIX are pursuing a linux compatibility layer strategy, which should allow Linux binaries to work out-of-the-box. I've not had the greatest success with OpenUNIX, but that doesn't necessarily reflect on Caldera. I guess you'll be buying support and hardware specific tuning (in AIX's case) while taking advantage of the masses of development done on and for Linux systems.
    • I don't know if it is on the binary level, but Sun touts linux compatibility as one of the new features of solaris 9 too.
    • HNot a Troll, really!

      Just want to know why you think this is not a reflection on Caldera. I would think it a balanced observation to claim that Caldera have been a poor organization in supporting others - even on community efforts that did not originate with them.

      Don't get started with SCO. The only nice thing to say about them is that they never put an OSF product on the market...

      • The ambiguity of text without intonation strikes again. I was just suggesting that it could be a PEBKAC error rather than anything wrong with the software. Not really a comment on Caldera at all. I will say that OpenUNIX seems to be fairly reliable compared to Open Deathtrap or Uselessware, as earlier products are affectionately known in these parts.
        • Got it!

          Don't mind me too much. I think of SCO, and I'm still sore that they made us pay extra to add a C compiler for SysV/386 - or whatever that beast was called. These days, that's standard commercial practice.

          No GNU C back then - we had 16 MHz 386's - Everex Step - that would compile stuff overnight with the SCO compiler. I even think this came from Intel, OEM.

  • I'm not sure. (Score:3, Funny)

    by tps12 ( 105590 ) on Friday September 20, 2002 @01:57PM (#4298574) Homepage Journal
    I think *BSD is dying, but beyond that I really couldn't say.

  • My extremely biased opinion (Score:5, Insightful)

    by photon317 ( 208409 ) on Friday September 20, 2002 @02:01PM (#4298608)

    I'm gonna go out on a limb here and be very opinionated and biased - but hey, it is an opinion question.

    I think Solaris will be the only commercial unix to survive more than a few short years from now (here I define survive as being useful, current and having a decent market share and attracting new users - technically Netware still survives today, but it hardly meets my definition, it's just slowly dying off as legacy users switch away).

    That being said, I don't think Solaris has much of a future either if they don't change their ways soon. They've already been trounced for web and application servers by thin cheap commodity linux stuff. They're only real foothold at the moment is large databases (think E10K-E15K class machines, 20TB databases, etc), and highly available databases at that.

    They're currently in the process of losing this to Oracle9's RAC linux clusters, which blow Sun away in terms of bang for the buck, and can scale just as well in overall bang and reliability.

    Now - all of the above is from the perspective of someone who believes only in the technical data and is willing to be on the slightly bleeding edge. When you factor in typical corporate environments/attitudes and whatnot, the picture slows down and pushes off into the future a bit further for the mainstream unix consumers.

    None the less, I think my assessment here will prove to be accurate over time, with all commercial unices eventually falling to Linux, with Sun being the longest and strongest holdout.

    Of course, in this long term sense, I really mean "the idea of Linux" when I say Linux. Linux could be supplanted by some other GPL (or GPL-ish) kernel down the line that re-uses a lot of the drivers and OS componentry from Linux and not really change my point.

    And for one final caveat, since I can't really see the future, I don't know for sure that Sun won't manage to correct their currently tragic course and get back in shape and survive. If they were smart, they'd stop trying to marginalize linux as a thin edge device, and start contributing large amounts of their man-hours to perfecting linux on the sparc64 platform as a way of protecting their hardware and support businesses against the demise of Solaris.
    • I despise Solaris - it's basically UNIX from 1987 with very few updates. That said, it is terribly important in many corporate environments, including mine (though I am now working with the applications, rather than directly with the OS).

      AIX is a fantastic UNIX, with lots of configurability and flexibility, running on quality hardware with outstanding support behind it. For some reason, not many people use it. I suspect that this is in part left over from the early 80s mis-steps by IBM and in part due to fewer applications being available for it than for, say, Solaris. But it's IBM, and it'll be around for a long time to come. It will likely outlast Solaris.

      MacOS X will be around for a long time to come, and if Apple expands their workgroup servers and enters the enterprise market, it could pose a real challenge to Linux's adoption in the corporate environment because it is valid on both the desktop and the server. Corporations like dealing with corporations, and this is an advantage to Apple.

      Most likely, though, MacOS X will continue to be a niche OS, and Linux will continue to be a niche OS (different niche, though) and Solaris and AIX will continue for quite a while. The other commercial unices are effectively dead letters in the long-term.

      • Fight! Fight! :)

        Yeah, Slowlaris does suck, but it has its good points, and I still think it's on top of the rest in the commercial market.

        Now AIX I have a bone to pick with. I've spent a lot of time with AIX in the 4.x versions, on a large number of RS6ks (SP/2 frames actually). I think AIX is a horrible implementation of unix, totally departing from the ideals of unix. I could go on for days on end (and I have, causing some of my co-workers to die of boredom), but to state a few simpler ones:

        Commandline text tools have arbitrary limits. AIX's built-in text tools (grep, sed, awk, etc) all have arbitrary limitations, like "no lines longer than 2048 chars". This flies in the face of the unix philosophy. Normally I wouldn't have noticed this as I don't regularly deal with lines that long. However, I regularly encountered such lines in the cluttered command output of AIX's configuration tool "smit", which is a whole other rant on it's own.

        System config data that should be in human (and unix tool) parseable config files in etc, is instead in a binary object oriented database. Further, the tools to manipulate the database are poorly documented if at all. Forget grepping or vi-ing your config. It's even worse specifically on AIX on SP/2s, all the SP/2 stuff is similarly in binary object databases.

        Their standard linker is really nonstandard and jacked, making configure and make scripts have many headaches.

        I have no opinion on MacOS X, except that I don't except it to be much of a datacenter contender.
        • AIX is a wonderful operating system. It's powerful, it's amazingly stable, it scales well.
          You just need to keep two things in mind:

          AIX is not Unix.

          Use SMIT for everything, or you'll blow your leg off.
          • Hmm... according to this [ibm.com] page, IBM is calling AIX a Unix operating system.

            I also know some places that teach Unix using AIX rather than Solaris or Linux.

          • AT&T is not Unix in the same sense that Linux or Solaris are. That is to say, AIX is not trying to stay very true to the original forms of Unix in every particular.

            AIX is, rather, a very enterprise-friendly Unix, with lots of nice disk management features, amazing hardware (RS6K) to run on, an object database for configuration (try reconfiguring your Solaris or Linux kernels without a reboot - OK, Linux can to some degree, with a performance penalty, with extensive use of modules) which is very smooth, and very high performance. As a former AIX admin (for years, including with the IBM AIX support group), and having been an admin also on many other unices (14 at last count), AIX is flat-out the easiest Unix to administer ever in an enterprise environment.

          • Using SMIT for everything becomes a problem at even a moderate number of nodes. The SP/2's I was admining at the time totalled about 45 nodes in dev/qa, and about 26 production nodes. Opening 45 windows on a control workstation and trying to simul-type in them (or worse, trying to use their auto-simul-typing cluster console thingy) was a nightmare, we had to go automated, which meant writing our own perl scripts to replace SMIT functionality. Since the AIX documentation is so poor, the way we had to learn was by capturing the commands SMIT runs (F6 or something? been a while), and basing our scripts on that.

            And yes, we blew our leg off a lot. A similar scripting approach has worked well for me in other unices, but AIX has far too many un-unixy idiosynchracies. Based on my experience, I don't really consider AIX to be a *nix, but rather a totally different OS from the ground up, that has had a veneer of slight unix compatibility applied.

            Their re-invention of several wheels has left them with a few more bugs than most. I still remember fighting rampant deep code bugs with their tech support, in subsystem ranging from their kerberos implementation, to their korn shell, to their resolver, and on. These are things that are somewhat more stable on other platforms that haven't deviated so far from the norm.

            • AIX is a *nix for all intents and purposes. It's extremely enterprise friendly.

              PSSP and the entire SP2 is an overexpensive solution that IBM has been toting for several years, and it looks like they finally wisened up. The idea of Scalable/Parallel is nice for the scientific community, but means less for real buisiness. Also noone from mainly a Unix background wants to fight both the SDR and ODM registry formats.

              I can't enumerate the number of times I've had issues with the SDR. Even with a complex built from scratch, it eventually turns to crap over time with multiple people managing it. The only way to avoid this scenario only exists in places like Lawrence Livermore, where procedure is preached like doctrine.

              The new Regatta's are slick. With redundant hardware that is built from the mainframe model. You want administration is is approaching the 9's of mainframe, it's going to come to AIX before you see it on Linux, mainly due to it being lightyears ahead already.

              From a buisiness model perspective, it's very simple.

              Windows = Cheap Hardware, Expensive Software
              AIX = Expensive Hardware, Cheap Software
              Linux = Cheap Hardware, Cheap Software

              The platform is only as good as the software that runs on it. You could have the cheapest, most robust platform that is always up, and well supported, but if there are no apps that run on it, then you have issues.

              • SDRSetObject alsdkgjlaskdj ladgfsldfhkglkrhe
                odmget alskdgflah jhfkghdfjhsdfg

                blah blah :)

                I'm sticking to my guns - I hate AIX religiously. Now on the other hand, I love IBM. They got some great engineering talent on the hardware and software sides, and I think they're going to go far with their linux strategies. I reconcile it by think of the AIX group as some rogue cancerous group within IBM that they're finally starting to take medication for.
            • would you not automate the administration of 45 individual unix servers? i don't get your point here - that smit didn't let you manage more than one machine?
              btw, every smit function is repeatable as either a command or a series of commands at the shell level. perhaps you need AIX administrators to manage your AIX machines?
        • They call AIX "Aches" for a reason. I started my foray into the corporate world supporting RS6ks and it has scarred me for life. It seems that IBM took the worst parts of System V and BSD, added a dash of mainframe riff raff, and blended it all together. A few things come to mind:

          ODM - Brings the dreaded M$ Registry concept to UNIX; Are you familiar with odmget and odmput?

          qconfig - Yet Another print queuing system for UNIX

          errrpt - Who needs syslog? Let's just make our own proprietary message logging system!

          Need to call support? Once you wade past the Level 1 morons reading from a screen, you better be prepared to take down your system and upgrade the microcode for every conceivable subcomponent, kiss those SLAs goodbye.

          Instead of using PAM, IBM came up with their own proprietary authenticaion API

          Startup scripts, who needs them? Instead of rc.local or /etc/rc?.d just put everyting in /etc/inittab. Better not edit the inittab directly, you have to use mkitab and rmitab.

          Man pages, you don't need those! Just to make things extra fun, we'll hide them on the Extras CD and instead of using troff/nroff, we'll put them in HTML format.

          let's-make-a-command-for-every-trivial-task; Put minimal info in the man page, and don't make the switches conform to normal UNIX semantics (e.g. -v == verbose)

          Instead of /opt for 3rd party packages, let's use /usr/lpp!

          Instead of dumping the information to the console on bootup, scare the hell out of the admin by making it appear the machine has hung. To humour him, print out esoteric hex codes on the LED panel. Wait at least 20 minutes before dumping *any* information out to the console!

          I could go on, but it brings back too many bad memories. I thank ${DEITY} for Solaris every day!

          • Damnit man, here i am trying to learn AIX for my job... and you're making me biased against it. (Okay, truth be told, i already didn't like it. The hardware it runs on is sweet in some cases, but the OS kind of makes me shudder.) I much prefer solaris.

            My least favorite is, like you said, the proprietary crap they've strung in, such as the lpp directory and errpt. AIX can swing for all i care.

            I had the experience of going to an IBM training course for WebSphere. The way IBM does things as a company is kind of strange.. i mean, dude, the CHAIRS had obscure numbering schemes. The hardware looked like it was TRYING to be 1980's back to the future looking crap.. it didn't look cool, and i dont think it ever did. Everything down to their phones - you don't dial 9 to get out, you dialed this three digit code.. To dial 911 you had to dial an 8 digit number to route through IBM's emergency services hotline. WTF?! Once you called them i wouldnt be surprised if you got questions like "Okay, the fire.. can you get the exact measurments and rate of spread? Okay, approximate temperature? Did you apply the -asbestos patch to the building? No? Well, then call back - we can't solve your problem until we go through the prereq of removing asbestos, which may indeed exacerbate the problem in this situation. Bye."

            IBM ERROR 12938-2391484

            They try too hard to be obscure.

            Being positive for a second.. Smit is nice.
            • If you are trying to learn AIX I would suggest you get The AIX Survival Guide by Andreas Siegert ... its available from Amazon et al.

              t
              • I'll look into it.. it wont spend a lot of time assuming i don't know what im doing in UNIX at all, will it? I mean, i'm pretty decent at solaris and sparc hardware.. but when it comes to IBM stuff, i dont even know hardware designations.
              • A nice book, but a wee out of date. It was published back in 1996 for christ's sake, when people still used 3.25 and 4.1 was newfangled.

                Since AIX is so fscking great, where are the books?
                • the same argument could be used that windows is much better than solaris, since there is so much more documentation available for it.
                  • Uhh, no.

                    My argument was that the only half decent book for AIX was written in 1996. Solaris, OTOH, continues to have well written books published continuously. If AIX is better, and IBM ships more UNIX hardware than Sun, one would think that the userbase would be larger and therefore there would be a huge market for AIX documentation. Supply and Demand, right?
          • Ok, someone needs to calm down a bit ... speaking as someone who has to look after a pile of AIX servers, (as well as HPUX, Solaris, SCO ...), ... I would address your points as follows:
            • The ODM ... yes it would be nice sometimes to be able to see things in plain text and be able to make a quick hack to edit them ... but the odm commands are not that difficult, (yes i've been there late at night using the odmdelete command!).
            • qconfig ... yep agree with this point.
            • errpt ... nothing wrong with errpt ... and you can always turn on syslogd if you really want syslog output as well.
            • microcode ... I disagree with this point ... the only time IBM support have asked me to upgrade microcode was due to a microcode problem (mismatch between adaptor and disk microcode).
            • I agree they should have used PAM.
            • ummm... the cd (marked Base Documentation) does have the man pages in the normal format as well as in HTML format. Yes you do have to install the seperately ... which means you do not have to install them on all your servers ...
            • Lots of small commands for everything - no problem with this ...
            • ummm... most 3rd party apps let you install them where you want, e.g. we install Domino in /lotus ...
            • Sorry but if your admin doesn't know that a RS6000 outputs codes on boot up then maybe he/she shouldn't be doing stuff on the servers, (e.g. knowing that sitting on 518 for a very long time on boot up is not fatal).

              You say in your last sentance you like Solaris ... fair enough ... I'm not particularly keen on it myself :)
          • the qdaemon system is a queueing system, not just the printing system. aix uses qdaemon to spool print jobs by default, and manage cron, and at, and user-definable batch queues as well. that said, it is still possible to use lpd on AIX, if you wanted to.

            syslog is still there - aix keeps hardware and software error messages separate from other messages, is all. just a different way of doing things.

            why would i need to call support? i learned the AIX commands to do AIX administration. my main application server is an RS6K running DB2 and CICS, accepting several hundred concurrent client connections per day, with an uptime of 326 days. administering AIX is not that hard, unless you insist on treating it like solaris or BSDI, which it isn't.

            i don't see where not using PAM is a liability, but if you're a shop where you need to use the same type of non-standard authentication across different types of unix, it could be more difficult than porting PAM modules between solaris, linux, and hpux, maybe.

            wow, all this time i've been using rc, rc.net, rc.tcpip to configure stuff during boot, but since it doesn't work, i better stop!

            no comment on the man pages :)

            since AIX has a package manager to add and remove software, why would it matter what the default directory is, as long as you know where it's at? if you install without using the package manager, there's nothing preventing you from editing the default .profile and adding /opt to the path, though.

            hahaha, no comment on the LCD :)

            I dunno. it's just different than solaris. that doesn't make it worse, though. it just means you don't get the job as an aix administrator.
      • damn, a pro-AIX post, and I didn't make it. what is this world coming to?
    • I'm not sure where it's going. I definatly see HP droping out of the comercial unix world over the next few months/years. The fact that they killed the project to migrate PHUX (sic) to itanium/merced, and killed the project for a new PA RISC chip (or whatever HP calles their unix chip) seems to prove this.

      Tru64 and alpha are dying. This is beyond sad. The EV7 is supposed to trounce everything in it's path. However, it's being canabalized by AMD and Intel.

      SCO/Caldera OpenServer et al have been decreasing in relevance for a while.

      AFAICT, that leaves Solaris and AIX.

      Solaris seems to have a good grip on the 'carrier grade' unix market. IBM is making a lot of good inroads here though.

      IBM seems to have the smartest strategy: "you know your pc hardware pretty well. Here's a linux on it. It's pretty cheap. Oh, want something more robust/fault tolerant? Give us some more cash, and we'll give you something that runs either AIX (with linux compat) or Linux proper. Want 5 9s? Here's our mainframe."

      That said, I disagree with madcalf (previous poster). AIX seems to do a unix stuff in a very non unix way. To be fair, I need to get better access to it... I havn't played with it recently. However, I'm confident that if I needed to, I could get pretty proficient at it in about 2 weeks of good hands on.

      To sum up, Solaris is entrenched, but seems to not know what it's doing stratagy wise. IBM is making inroads, and has a solid stratagy, but it's the underdog, and IBM has never been able to sustain a marketing drive. Everyone else is dying.
      • There's also BSDI, which seems to be doing well in their niche. The model of working with a closely-related open source Unix seems to be working for them - I've seen lots of products based on BSDI (like proprietary IDSen and firewalls).
        • I was mostly focusing on the "comercial unix out of the box" space. The "appliance" space is a different ball of wax. BSDI does well there, as does FreeBSD and Linux. All three can be hardened to the gills, and you have all the code for 2 of them (and most of the code for the third).

          I really think that the free unicies are going to continue to dominate the 'apliance' space. And I think the use of them will grow greatly over the next few years.
      • I'm not sure where it's going. I definatly see HP droping out of the comercial unix world over the next few months/years. The fact that they killed the project to migrate PHUX (sic) to itanium/merced, and killed the project for a new PA RISC chip (or whatever HP calles their unix chip) seems to prove this.
        What are you talking about? HP-UX for IPF isn't just a project, it's a product [hp.com].

        Note the part about Linus binary compatability by next year -- yet another sign of the Tru64 influence in the process.

    • Accordign to Netcraft and IDC, Red Hat have more market share than every other Linux distro combined - I think commercial Unix is doing quite well, as long as the business model doesn't involve proprietary software.

      Unless you didn't mean commercial Unix and instead meant proprietary Unix which I think has a limited lifespan and will be relegated to the truly high end for the next five of six years as (likely commercial) Linux erodes their market from the bottom end up.

  • I just do not see Sun and SGI going backwards in order to run Linux on their proprietary mid to high-end systems. Why would they abandon or por resources into duplicating what they already have?

    That said, I do see Linux and perhaps *BSD becoming more prominent running on new hardware from the above and a couple of other vendors.
    • I just do not see Sun and SGI going backwards in order to run Linux on their proprietary mid to high-end systems.

      I'm sorry to say that SGI is, in fact, running Linux on their high-end systems. The decision was made several years ago not to port IRIX to the IA-64 architecture, but to instead work to get Linux to the point where it could be usable in a large-scale technical computing environment.

      By all reports, expect to see very large-scale SGI servers based on the Itanium 2 to be released either this winter or next spring with, yeah, a Linux kernel.
      • why would sgi want to put IRIX on intel? intel hardware isn't impressive at all for i/o intensive loads. the cpus are pretty fast, sure, but stick that with a workstation-derived memory controller and bus controller, and you've got major bottlenecks. obviously not an issue if you're running a renderfarm on them, true enough. i hear IBM is doing some interesting stuff with smp servers, but i've tested compaq 4 and 8-way servers, and they're readily beat by my 5-year-old 4-way rs6k in terms of memory throughput and i/os per second.
        • You don't get it. SGI wasn't talking about porting IRIX to PCs. They were talking about porting IRIX to Intel architecture CPUs, and then putting those CPUs in Origin-series servers.

          This is, in fact, basically what they've done. The new product line coming in January is an evolution of the Origin system architecture that can support either MIPS or IA-64 (in this case, McKinley [Itanium 2]) CPUs, with IRIX on the MIPS systems and Linux on the IA-64 systems. Whether they're going to ship both MIPS and IA-64 systems or just IA-64 systems is unclear at this time, but the prevailing opinion is that you'll be able to buy either. Same architecture, different CPU, different OS.
          • got it now... so when this goes down, will the linux systems sold by sgi run interactive desktop/4wm/whatever, or will they be using gnome or kde? will they release cosmo and annotator and the rest of their tools for it?
            • I doubt very seriously that SGI's servers will come with either Gnome or KDE. These machines-- multi-hundred-thousand dollar systems that start with 32 Itanium 2 processors and scale up from there-- aren't meant for interactive use. They're multiuser server systems. That said, SGI might include an X server, or even a desktop environment, just pro forma.
              • well, i thought that, given the power of the origin servers (even the 300 line) it might make sense to run one or two origins and several Xterms instead of putting an O2 or Fire on every desktop. This is why i was wondering about ID and the tools... wow, this could be a big coup for SGI, since i'd think they'd have gotten the intel chips to work with the numaflex shared memory setup. that could be cool indeed.
    • SGI's IRIX is slowly going away. I'd have to disagree with your statement about them. I think that they are going to be dropping IRIX in the future. I also wonder if they are doing any development on IRIX or just supporting their existing client base.

      Sun on the other hand will probably stay around for a while as will HP.

      AIX is slowly going away.

      Mac just released OS X.2 and while it is BSD and not really UNIX it is UNIX like enough to be considered something that is not going away. The other BSD (FreeBSD, NetBSD, OpenBSD) will probably stay around as well because they are open source and someone will want them for something. They are all pretty good in one way or another.

      Linux has a few more years to go before it 'stabalizes' into a less of a moving target. Code that worked on 2.0 may or may not work on 2.2 or 2.4. This lack of backward compatiblity is one of the big problems many people I have talked have with Linux. Which version do I port to and how much maintenace do I have to do. Then how do I support all the 'flavors'? What works on SuSE may or may not work on RedHat or Slackware or debian.

      1) Linux needs to become and move 1 good package mangerment system. RPM is okay, but not good enough. It should be smart enought to get and resolve all dependancies from not only what is installed through rpm, but also what is not installed through rpm, much like BSD ports system works. Also one should be able to remove the package and its dependancies if they so desire without breaking the system. I think apt-get does this, but never used it. I know BSD system can.

      2) GUI desktop. While KDE and GNOME are great in their own right, they really need to be only 1. End users don't care that much about the GUI as long as it is easy and User Friendly. Look at cell phones and pda these days. They are for the most part fairly easy to use. I really think the desktop as we know it will disappear in the not so distant future. It will become more user friendly. People wont program with a text editor they will have a gui that generates ALL the code for them. If they have a file they will describe the format of the file and the editor wil deal with the fopen(), fread(), structure declaration, and fclose() and all that crap for them efficiently. You'll get home touch the screen on the computer and it will get your mail or check your appointments. You will have the option of writing, or using an onscreen keyboard, or voice to enter in data. While I know that there are people here that love the command line, there are more that don't. Why do you think that we have GUI in the first place? So that people can do things easier.

      I'd say 5 years and we will see many changes.

      • Mac just released OS X.2 and while it is BSD and not really UNIX

        Who's Mac? I think you mean Apple (or really, NeXT). BSD is very much pure UNIX, it's Berkeley's fork of AT&T's UNIX from long time ago. BSD has been a major influence on "modern" unices, take a look at some of the UNIX family trees / timelines. Though SunOS 5.x (Solaris 2.5 and newer) has a lot of AT&T System V in it these days, it was once almost pure BSD... as were most west coast flavors of UNIX.

        IRIX is alive and kicking, it gets a major overhaul every quarter, as does most of its major subcomponents. Over the past 24 months, revisions of IRIX 6.5 have rolled in seamless support for SN2MIPS (O3K/O300/Fuel), VPro/Odyssey graphics, InfiniteReality3&4 graphics, as well as a major compiler and runtime upgrade (MIPSpro 7.3), clustered/distrbuted filesystem (CXFS), and gobs of media and gl libraries. The help system was recently totally overhauled, CXFS continues to get major upgrades (and more cross platform support). Recently announced were still more major overhauls... MIPSpro compilers will be upgraded again in October, Performer is getting a huge upgrade by the end of the year, Java is being beefed up, the freeware.sgi.com archive is growing. Plus there are a lot of things that just work... I wish other vendors had *half* of the performance profiling tools that IRIX includes. SGI kit isn't cheap, and parts of it aren't too flashy, but it's hard to beat if you need the torque.

        The version numbers don't increment much, but there's been a lot of behind-the-scenes work and gobs of improvements addressed across the board. The rest of the industry has been moving so fast that one no longer needs an SGI to do most 3D or video work, but should you need several pipes of graphics for a simulator or need to shuffle several streams of uncompressed 1080i, then there's no better platform. It's a tough industry, and SGI's financials suck, but they're not doing too bad.

        Yeah, I wish SGI would update their desktop a bit more, but then I also wish Cray would implement some major changes in UNICOS userland too... but it looks like I'll have to wait for both.
        • "BSD is very much pure UNIX"

          WRONG WRONG WRONG!!!!! BSD started out as an add onn to UNIX. From an article that is still hopefully here -> http://www.extremetech.com/article2/0,3973,555451, 00.asp

          "Here are the origins of BSD and the operating systems it has spawned. BSD stands for "Berkeley Software Distribution," the name first given to the University of California at Berkeley's own toolkit of enhancements for the UNIX operating system. Created by the students and faculty, BSD was not part of UNIX itself, which was created by Bell Labs. Rather, it was a widely distributed package of software enhancements for UNIX -- a supplement that made the operating system, which was originally strictly a research vehicle, useful in the real world"

          It is therefore very much like UNIX but it is not UNIX.

    • Why would they abandon or por resources into duplicating what they already have?

      I can't speak for Sun but where do you think Linux got OpenGL and XFS from? SGI most certainly is spending time and money porting their stuff to Linux. From their perspective if they can port over enough stuff they can get out of the OS business and focus on hardware (where they can make money).

  • HPUX's future (Score:2, Informative)

    by Blob Pet ( 86206 )
    My understanding is that components from Tru64 will be integrated into HPUX in future releases. This article [hp.com] states, "HP Tru64 UNIX, an integral part of HP's UNIX portfolio, also showed strong results in the DH Brown report, earning top marks in two of the report's five categories. As part of its ongoing plan to continue enhancing the functionality of HP-UX, HP plans to bring key features from Tru64, including its TruClusters and Advanced File System capabilities, into future releases of HP-UX11i."

  • Very humble HP-UX, DG/UX, AIX opinion (Score:3, Interesting)

    by Bravo_Two_Zero ( 516479 ) on Friday September 20, 2002 @02:40PM (#4298916)
    IBM seems the most likely (having publicly stated as fact, at least) to give up their commercial *nix in favor of Linux. But, as good as Linux is (and I have it everywhere), it still has maybe a decade to go to have the sort of stable lineage of the other commercial *nix products.

    That isn't to say it's less stable, exactly. It's just a matter of how many times vital features have been tested, added and changed. HP-UX, as an example, has a bright future. HP is integrating features from Tru64 (volume management, etc) to an already supremely stable platform.

    By comparison, DG/UX (Data General) still has more advanced features (NUMA, NUMA, NUMA, etc and did I mention NUMA? HP doens't want to listen on that one), and development died on DG/UX years ago. But HP-UX soldiers on because of stability and compatibility with large installations. Heck, even the old K's are going for a premium now because some customers won't give them up.

    I'd even suggest Solaris might be in the same boat. The install base is huge. The reliability is outstanding. Linux and BSD can compete, but they have a ways to go to get that install base.

    I'd say the future is less bright for non-*nix commercial operating systems. Netware has the user base, but I don't think it's grown in a few years. NT/2K is growing, but companies won't stomach the restrictive and expensive licensing for long. Forget about the promise of .NET... it's the price tag that will kill it.
  • I talk to a lot of old school Unix guys (who prefer Solaris for it's Unix pureness). They seem to agree that the desktop isn't important. I think they're wrong here though. The future of any Unix is tied to the desktop.

    Microsoft first controlled the desktop operating system. Then, they used the desktop operating system to control desktop applications. Then they used desktop applications to control network services/mid-level servers. (Outlook/Exchange, IE/IIS). Now, they are trying to strengthen their hold on mid-level servers and break into the high-level applications department. (Not through high-level servers but a bunch of mid-level ones).

    They may or may not ever get the high-level servers, but I think if history proves anything, they will eventually. Not because they have a better product, but because they are able to use existing dominations to leverage into new markets.

    Therefore, unless Unix makes headway on the desktop, there's no way they'll servive as servers. This is why there is hope in Linux as a desktop. Without it, Unix will eventually die.
    • The desktop isn't important in a server enviroment. Or at least as important as you make it out to be. Saying the fate of unix lies in desktop doesn't make sense when unix has never been geared for mainstream desktop use.

      What is important is interoperability with desktops, and more importantly desktop applications.

      • Saying the fate of unix lies in desktop doesn't make sense when unix has never been geared for mainstream desktop use.

        this statement is just so wrong. What do you think the desktops of the internet looked like before Win9x and NT3.5 were in widespread use?

        Also, the rise in use of MS for servers has corresponded pretty well with their takeover of the 'desktop' / 'workstation' market and hence the decline in use of Unix as a workstation.

        Further, if one would care to consider a VT as a desktop of sorts - one that is unix if connected to a unix host, it becomes even more true that the fate, and late 90s decline, of unix is tied to the desktop and the decline of use of Unix on the desktop.

        If you have a bunch of Sun machines as your workstations, you're likely to have a Sun as your server, (substitute Sun with SGI or DEC Alpha, etc..)

        --paulj
  • I see commercial unixes continuing to thrive for at least a few more years. Linux still isn't mature enough in the opinions of many corporations and sysadmins to handle the workload. Sure, the kernel support, etc is there, but until the Itanium becomes mainstream and middle ground software vendors begin supporting, en-masse, linux or another free operating system, the enterprise will still be overall lukewarm.

    The Itanium will be a very positive thing for Linux, as it will likely be the Intel unix that replaces SCO and DG-UX's niche. There's another whole topic that can be written about the future of commercial unix processors.. AIX and HPUX run on Itanium natively. Solaris may or may not (they got it to run back in '99). The SPARC processor, long one of the leading hardware architectures, is going to suffer to the EPIC/VLIW capabilities of the Itanium, from what I can tell. The future looks grim right now, but don't count commercial vendors out yet. I'm sure there are a few tricks up sun's sleeve - they're already looking towards preliminary designs for the sparc VI and VII, and multicore sparc chips.

    This is the kind of argument i dislike getting involved in, but i felt compelled somehow to put my $0.02 in. Being an objectivist on slashdot is akin to being a gay atheist in a Baptist church. You either get converted or killed, and most people don't bother with conversion anymore.
  • Well, this may not be what you expected but MacOS X seems to being well. It's clearly a Unix, is now on more desktops then any other commercial Unix, looking to move further into servers, etc.

    • OS X is not a Unix.

      Darwin is a Unix. But that doesn't make OS X a Unix.

      Now, if OS X has X11 installed on it (or no GUI at all...) then it can be called a Unix. How many OS X installs can you say that about?

      • Re:MacOS X (Score:4, Insightful)

        by inburito ( 89603 ) on Friday September 20, 2002 @04:23PM (#4299598)
        Phuuh.. Because it doesn't run X11 as default you say it is not a unix! How did a windowing system suddenly become unix?

        Technically when you're talking about unix you're referring to a trademarked owned by Open Group but practically we mean posix compliance and os x does a pretty decent job at this.

        It doesn't matter if it runs by a means of a sledgehammer as an output device and a chess table as an input device as long as it otherwise conforms to posix specification (api, shell, utilities) you can for all practical purposes call it a unix..
        • Technically when you're talking about unix you're referring to a trademarked owned by Open Group but practically we mean posix compliance and os x does a pretty decent job at this.

          isn't windows NT (2k, xp) also posix compliant? I seem to remember that at least NT4 was, since the us government (possibly the military) will only buy posix systems.

          if so, what reasonable server-os is *not* a commercial unix?

          • NT/2000 implemented some portions of the POSIX standard. It left out some fairly significant things like POSIX threads, signals, etc...

            A little bird said that .net would not include the POSIX support layer anymore.
        • Technically when you're talking about unix you're referring to a trademarked owned by Open Group but practically we mean posix compliance and os x does a pretty decent job at this.

          We're giving this blowhard kid waaay too much attention but actually MacOS X is a legitimate user of the Unix trademark. The Open Group did a press release about it awhile ago and you can see Apple listed here [unix-systems.org]

          MacOS X is Unix. It is technically. It is legally. It is functionially. Indeed it's the best-selling commercial Unix out there.

          That some poor dork confused GUI or lack of or X Windows or whatever as "Unix" is between he and his doubtless depressed compsci faculty.

      • OS X is not a Unix.
        According to the folks who hold the Unix trademark it is.
        Darwin is a Unix. But that doesn't make OS X a Unix.
        Darwin is different because...
        Now, if OS X has X11 installed on it (or no GUI at all...) then it can be called a Unix.
        Uhm - what fucked up universe do you come from? Do you even have a clue what you're babbling about?

        Dismissed. Go get some education amnd come back if it sticks.

      • So you are saying that unix didn't exist until the mid-80s? riiiight.
      • Darwin, the BSD fork that OSX is built upon, is AFAIK developed openly and independently from the Aqua GUI. I have not tried it but you can download XFree86 for Darwin at osxgnu.org [osxgnu.org]
        • I have tried, I do use it (though not the OSXGNU build) and it works fine rootless and rooted. In full screen X is a doc icon so you click on the X icon to go to the X screens and use a keyboard combo to go back to aqua. Rootless you just click on the windows and X even has .nib file so you get an "apple menu" with your rootless X apps.

          I've also used the client against emulated dumb X terms and it works perfectly.
      • I don't buy it. First off there were many Unixes that didn't ship with X11 for many years; X11 owning all of the Unix gui market is fairly recent. Further all it takes is:
        fink install xfree86-rootles

        The way I see it its closer to feeling like a Unix then many other Unixes I've been on. It ships with perl, gcc, ...

  • I think people get slightly over excited about the features of linux. Yes it's good, yes it runs on lots of hardware, and yes it even runs on quite big iron.

    But, until the software support for linux is at the level it is on IRIX you won't see anything. One of the main reasons we keep our 64bit IRIX boxen about is the visualization software that we simply cannot get for linux. I'd also wait quite a while for 64bit mainstream processors settle before I'd expect linux to make much of an inroad into that market.

    There really isn't much of a move on the software front, even still. We'd prefer to be a linux only shop, mainly for support reasons, but it's nowhere near possible at the moment.

    The level of support we've received from SGI has also been nothing short of first rate. The cost isn't really that much of a concern, so linux doesn't really attract us from that angle. An open-source OS would be a bonus though (we're an R&D shop, so transparency is great).

    Give it time...


  • LiBSDnuxFree will be the dominant commercial Unix.
    Everybody in unix is sharing code like some big frat party in the land of loose women.

    Open Source has changed the rules of the game in this respect.

    OS X, probably the highest volume commercial license out there will continue for the next decade, and all the variants that currently exist will continue to exist-- in name only if they're proprietary-- and in caode and many more flavors if they aren't.

    Operating systems are becoming pretty much a branding opportunity. How much of Solaris and OS X are in common? At the OS level, a fair bit.

    So, everyone keeps current, everyone shares code, everyone keeps their "proprietary" brand and sells their stuff, plus whatever value add they bring to the table.

    The real question is how long does Windows have to live? Probably the next ten years for sure, but forever?

    Really, with OS X its become Microsoft on one side and everyone else on the other... and there are more smart people combined working on/for Red Hat, Apple, IBM, Sun, HP, Debian/Linux, FSF, and all the others I'm sure someone will remind me of.

    No matter how many people MS hires, most of the brains won't work for them.

    Sure, there will be variations, but not in what's important. For instance, say the Debian project comes up with a killer filesystem that nobody else has, and its open source (of course) then when it becomes a competitive advantage for Debian, Apple will adopt it, IBM will adopt it, Sun will adopt it. Suddenly, not just Debian is working on this filesystem, everyone is. (You can switch any of those players for Debian. IF apple came up with the new killer filesystem, then everyone would eventually support it because Darwin is open sourced. Its only a matter of time, and the fact that it isn't compelling, that Linux (for instance) doesn't support HFS+ better than it does.

    So, with everyone united behind the OS Unix codebase, they will keep selling their brands and their value adds, but all the competitive advantage will be shared. And united against Microsoft.

    In a way, all of them have gone away as commercial unixes, and in a way, all of them are thriving.

  • I haven't been following any IRIX roadmaps lately, but it seems that IRIX 7 will likely never happen. I went to SGI's Developer Forum in '97, when 6.5 was relatively new. Several of the sessions talked about all the great things that they were doing for IRIX 7, due out no later than the following summer. 5 years later and we're on something like 6.5.17. SGI continues to pump out minor maintenance and feature releases, but you certainly get the feeling that IRIX isn't going much further these days.

    'Tis a shame, really. Back in the day, IRIX was the best Unix out there, bar none (my opinion, of course). It has some fantastic technologies that still aren't matched in most other Unices (guaranteed rate realtime I/O - would love to see this in OS X, NUMA, hundreds of processors running off a single system image, OpenGL integration, more) and had the best/easiest user interface and admin tools at the time. Sadly, SGI has struggled the past 5 years and has done little of interest with their OS. As others here already said, I recall reading that they're working on adding some of their high scalability technology to Linux so it will run on their big iron. If this is their current direction, R.I.P. IRIX.

    I still like my O2 at work, but now I'm just bummed that my request to replace it with a new PowerMac was denied. OS X rocks.

  • Why would anybody use a proprietary unix when Linux has the following advantages:
    • Cheaper
    • Runs on a wide array of hardware
    • Is rapidly evolving
    • Supports the latest hardware gadgets
    ?

    Well, the reason people in some cases will proprietary unixes is because the above advantages are not free or does not matter. Eg:

    • Cost. When you also take into account the hardware, training, saleries, the low (zero) cost of the OS is not that much.
    • Vendors do not care if a modification/optimization will not work on a peice of hardware they do not support. "Ohh this optimization will break on Pentium3" - the vendor does not care if he does not support Pentium3 and is thus able to do some optimization that Linux cannot. Vendors are also more willing to create two versions for low-end and high-end hardware.
    • Linux is evovlving. The flipside is that it not stable (as compared to proprietary unixes). With a proprietare unix you get a release which has been tested. And you receive patches. And documentation on the patches. For a system administrator this is important. It is also important for ISVs who are sometimes asked to guarantee that their software/hardword works on with customers hardware/software. It is much easier to guarantee for a stable prorietary unix than for a Linux system with unknown/custom kernel.
    • Support of legacy hardware. Vendors will usually do much to support legacy hardware. Linux can afford to say that an old piece of hardware is junk and will not be supported. A vendor cannot do that.
    Sometimes a proprietary unix can support specialized hardware sooner. Eg. assume that a vendor is developing a WizBangUltraBus which can automatically page out memory to disk, directly interact with the onboard DSP, act as a hardware watchdog, and lots of other things. It is much easier to add support for that in the vendor's proprietary unix than spending months convincing the linux-kernel-mailinglist that support should be added.

    Proprietary unixes also has some features that do not exist (yet) for Linux. High-performance distributed lock manager in cluster systems? I would use TruCluster DLM. Want nice administration of a bunch of machines? I would use AIX's DSMIT. Want native support for Veritas filesystem? Try HP-UX. Want LPAR? Try Solaris or HP-UX.

    There are also a few other issues such as support for hardware with NDAs. There is also the CYA factor.

    • I forgot to write something about the future of proprietary unixes. Another poster's earlier comment [slashdot.org] wraps that up quite nicely.
    • Just a quick note - most hardware is getting so cheap that it's cheaper to replace the hardware if it isn't supported. Think scanners, printers, cd-roms, worm drives, flash card readers, removable storage, hard disks, memory sticks, - if it isn't supported, give it to some poor sucker you don't like and let them waste time with it.

      Most periperals, heck, most motherboards and cpus, are becoming "disposable" in the cost equation.

      Besides, if you want support, you're always free to either: (a) add it yourself or (b) pay / bribe / cajole someone else to add it. This IS one of the benefits of open source, after all.

  • I think its important to take a question like this and break into the 5 main markets:

    1) Home user / small business
    2) Corporate desktop
    3) Departmental server (large business)
    4) Enterprise class server
    5) Embedded system

    Home / small business is territory firmly under Microsoft's domain. Historically none of the Unixes have penetrated area (1) very much. In terms of Unixes Linux and OSX are the only Unixes even trying. IMHO they are likely to separate on price. Since the death of Commodore the under $500 market has been wide open and Linux has a chance to make a real entry at this price point, where the cost of Windows XP is a real factor. At the higher price points, IMHO OSX offers the more compelling package with OSX being a much easier to use desktop and OSX Server being a very inexpensive and easy administer small business server.

    In terms of the corporate desktop the death of terminals and the rise of the PC desktop established Microsoft firmly in area as well. This market is also cost sensitive and here the cost is not only Windows XP Professional but also Office. The real question comes down to whether the cost of Linux desktop support eats up the cost savings on the software. Sun has wanted this market badly but has never offered a product that interests the corporate desktop market. Outside of Sun I don't see any of the commercial Unixes even trying. The is one odd possibility that United Linux is very successful and you end with a commercial Linux taking a good chunk of this market.

    In terms of departmental servers Linux is really shinning. More and more the x86 hardware is able to handle most low-medium loads comfortably and administrators familiar with Linux are readily available. The fact that Linux offers one of the best "working environments" of any Unix means IMHO they could easily end up owning this environment outright. The real competition is Microsoft's server line with its:
    1) GUI that works off the familiar Windows GUI
    2) SQL Server / Access line up for ease of development and deployment
    3) Active Directory providing real added value over DNS
    4) Its already strong presence
    I don't see much that the other Unixes have that Linux doesn't.

    In terms of Enterprise class servers the nice thing is that Microsoft isn't a player.
    I think its worth breaking this into two sub-areas:
    a) Supercomputing
    b) High reliability computing
    In terms of supercomputing Linux is actually doing quite well. Between IBM and SGI working with expensive hardware and Beowulf working with lots of cheap hardware Linux is making a strong showing. The strongest competition here is AIX and IRIX and since they are made by IBM and SGI respectively Linux's odds look quite good.
    In terms of High reliability computing Linux really hasn't made a strong play. This is the stronghold for Solaris, HP-UX, AIX.... I don't know if open source methods end up working here, the desire for rapid improvement and incredible boring tedious work required to get from 3 9s to 5 9s might mean that Linux can't win here. To be honest though I really question whether Unixes should be here at all. This seems like a much better place for systems designed with higher security from the ground up: Z-OS, VMS, EROS, etc... IMHO should own this space though they seem to be losing ground to the Unixes.

    In terms of Embedded systems I think its early to call. Thankfully Windows CE is not a success so this area has remained competitive. Here I'm not even sure a Unix will end up winning. QNX has a great product and this is sort of Unixy. Palm has bought BeOS.... Linux is popular. I have no idea how this one plays out.

Slashdot Top Deals

I've noticed several design suggestions in your code.

Close