Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?

IBM banks on Linux 185

jdaily writes "IBM's server group head said in an interview that IBM will Linux-enable all of its server hardware, from PCs to mainframes. " This is a pretty major endorsement... but I still want a Thinkpad running Linux with every component (including the freakin' modem!) working. You listening IBM?
This discussion has been archived. No new comments can be posted.

IBM banks on Linux

Comments Filter:
  • by Anonymous Coward
    It was kind of gutsy for you to lay the truth right out there on the line in your aritcle summary, Rob. The truth is that IBM and Compaq and most of the 'bigs' now "Championing" linux have little or no enthusiasm for the Linux desktop. Linux gets endorsements as a server platform, and somewhat for embedded designs where it's no more visible to the user than the firmware in a Microwave oven. Nobody in the commerical world (except for the feeble offerings of shrink-wrapped boxes with Linux on them) endorses Linux as a general desktop solution.

    To my mind the Linux 'desktop' has only gotten uglier over the past year as more and more of the clean design implied in an OS that can be administered using vi to edit text configuration files. Linuxconf sucks bigtime. Gnome and KDE have turned into twisty little passages going every which way.

    Applications I used to like using have fallen by the wayside, because they didn't have K for the first letter in their title. I see a sea of the kind of apps that college students can throw together on a weekend, but little else for the desktop. Corel drags out the tired old WordPerfect zombie, but NONE of the vector-based drawing tools that they made their name with have been ported, and it sounds like they're scrambling with the Wine kludge as their porting strategy.

    What a pity, but I guess it's what one can expect from a project with little or no design foresight. I have NEVER seen anybody involved in Human-factors design involved in a single Linux GUI project.

    Oh well. Don't cripple your Thinkpad yet by installing Linux on it.
  • Er... IBM *owns* Lotus. All IBM has to do is say "Oi! Lotus! Where's my Linux client??? I want it NOW" - and Lotus has to do it. So just get enough IBM staff who want to use Linux together, and tell Lotus what to do. Route it thorugh IBM management if you have to, but if you have more than 300-400 employees shouting for it, management will pretty much have to comply, or lose thier skills base.
  • by Anonymous Coward
    Sure there is. It's called "Netscape" and it works quite well, thank you!
  • ... on my Thinkpad 760E. The XFree86/SVGALIB support is perfect (full use of all modes). Both PCMCIA slots work fine. The IBM EtherJet card works (in a fashion). With a little mucking about IrDA works (for syncing to PalmPilot). Floppy drive works fine. APM seems to work sometimes.

    In fact the only damn thing which doesn't work is the MWave modem and sound. Using the "DOS Hack" you can upload soundblaster 1.5 emulation code to the MWave DSP, then the standard Linux soundblaster driver works. But you can imagine that booting through DOS/loadlin isn't a lot fun.

    One incredibly simple fix that only IBM can provide would be to release the source code to MWD. IBM already provides a good collection of precompiled MWave/DSP code for the benefit of DOS users. With MWD/Linux you could upload these DSP binaries and then use existing Linux drivers with the emulated soundblaster 1.5.

    At least then us Linux/MWave users would have sound, and I can personally live without the builtin MWave modem, but IBM doesn't even bother to reply when you politely ask for assistance (customer support, yeah right).
  • I believe an open, LINUX OS on the mainframe would make software portability easier, and would make the argument to migrate both the software and protocol stack to a pure IP environment much more palatable from a business perspective.

    I dont know if you are aware of this, but there already is a version of UNIX running on MVS, and it has been around for a few years now. Its called MVS/Open Edition, and the API is quite public. Any portabilty issues could (in theory) be addressed on that platform.

    However, IBM's Linux strategy does make a lot of sense. As you correctly pointed out, the portability will indeed improve, if what they are trying to do is port a Linux application from one platform to another Linux platform.

  • I, too, am a proud owner of a ThinkPad 770 (no E/ED/X/Z) 9549-1AU. It does have the MWave modem, but I don't use it (using LinkSys CardBus 10/100 card).

    Most of the items on your wish list are already availible. My 770 runs practically flawlessly. I have accelerated X (which works pretty well with a slightly modified XF86_SVGA server; more on that in a minute), fully functioning APM (suspend and even hibernate to disk), PCMCIA and CardBus, and sound.

    I don't know which XF86 you're using (I'm on an older one, 3.3.3 I think; I'm not at the laptop right now), but I had trouble getting the SVGA server to run accelerated, even though it claimed that the Cyber9397 was supported (I'm using the first release to claim support for acceleration on the 9397). I did a little research, and found this post [] on which describes how to modify the Trident server source code to enable full acceleration. My understanding is that later versions of XFree86 do not require this modification, but I'm not sure since I haven't had to upgrade mine for a while, and so I consequently haven't bothered. Also, read the rest of the messages on that thread; after making these modifications, you will need to add the options "tgui_pci_read_on" and "tgui_pci_write_on" to your XF86Config otherwise the server will hang.

    Also, I used to have problems with X and APM, such as when closing the lid or suspending the machine while in X, the screen would get really funky when waking it up. For some odd reason, these problems cleared up after I built myself a kernel with the VESA framebuffer console driver and then started using the console at 1024x768x16bit. Now X gives me no problems, but only as long as I'm using the VESA console. Perhaps the VESA console drivers do something to initialize the graphics chipset that the SVGA server fails to do? I don't know. Using the VESA console also fixed another problem I was having with the accelerated server: before using the VESA console driver, I was having trouble playing back sound (MP3) in the background while I was working on something else. The music would skip if I so much as moved a window. After doing some research I concluded that the Trident driver must have the "pci_retry" option permanently on by default. I could find no way to disable it. Strangely enough, the VESA console solved this problem too!

    And finally, you might want to check out tpctl for Linux [], which is a program similar to the PS2 program for DOS that comes preloaded on ThinkPads. It allows you to control various things in the ThinkPads with a SMAPI BIOS. And the latest version even comes with a patch to hdparm that allows one to hotswap their IDE devices in the UltraBay! I haven't had a chance to check it out yet myself, so I can't tell you how well it works, but I thought you might like to know. :-)

    Feel free to e-mail me if you have any questions.

  • Oh, and BTW, I'm running Linux kernel 2.2.10 (with the suggested PCMCIA services version that was listed in the Changes file in Documentation/) on Slackware 3.5 (yeah, I know it's old, but it works great, and if it ain't broke...)

  • Recent ThinkPads (the 770 comes to mind) use the MWave DSP only for the modem functions, and not for sound. My 770 has a Crystal chipset, too; the 4237, which works flawlessly with the Crystal 4232 driver that comes with the kernel (except for the fact that it doesn't support the SRS 3-D control in the mixer).

  • Some people have talked about the possibility of Linux and the AS/400 playing nice. Man, that would be nice. Rumor was about a year ago that IBM had already ported Linux to the 400 native, and if that's true, I want it.

    But frankly, at this point, it's not the best thing IBM could do for the AS/400 and Linux. OS/400 is a better match, since the AS/400 is such an odd box. Making OS/400 gratis would be a better step then getting a good port of Linux to the 400.

    But we all know that's not going to happen. I don't see why IBM doesn't instead focus on making it so that people would want to use the AS/400 more. Namely, work towards getting the AS/400 to integrate a little nicer with Unix. For instance, telnetting from an AS/400 to a Linux box is full of trouble - passwords aren't obscured, Pine, Lynx is unusable, etc. If they could just make it so that there was something along the lines of (although probably not specifically) a TN5250 termcap entry and maybe a 5250 getty or a 5250 API library for unix that would allow at least basic vt100-level compatability. Being able to host AS/400 applications off a unix box would be great for AS/400 ASPs to develop software for AS/400 users without hiring AS/400 developers.

    Also, it would be good if IBM could get an Apache port to the 400, and make it so you can download and compile it from Apache on the AS/400 with modperl would be a godsend. BTW - does anybody know if this exists? If so, please let me know!

    It would also be nice if IBM would pump some money and a programmer or two into the Linux TN5250 client - it's so damn close, but not yet production level. The problem is with closed protocols, it's hard to do. Why doesn't IBM focus there?

    Again, IBM, we don't need an AS/400 Linux port right now, we just need OS/400 to be more Unix and Internet friendly. Please.
  • R5 saw the end of the Unix client development stream. The only option you're going to have is either Wine or HTML based clients. Even the Java-based security projects are so specific to Wintel that it's useless to try to run it on *nix. You're not going to see a push for a Linux/Solaris/AIX/HPUX client with the current goals and you're not going to see a lot of browser interface development due to the Wintel people still having a non-browser Notes client. There are a lot of *nix projects internally that have picked up the Linux banner on the server side and there needs to be a groundswell of non-Microsoft client requests before they take notice. They make money selling servers, browsers are free... where do you spend your development dollars?
  • As an IBMer who has worked through the legal, marketing, and executive teams to get commit status to PHP [] and Jakarta [] let assure you that it can be done.

    And none of this would have been possible without the pioneers who blazed the path towards open source within IBM with contributions like Jikes.

  • I have NEVER seen anybody involved in Human-factors design involved in a single Linux GUI project.

    I guess I agree with you about that. If we want the average computer user to put Linux on their desktop we have some work to do.

    I think when or if real human-factor people do work on the the different GUI's available for Linux (insert your fav. unix-style open source os here) is when it will go big time with regular end-users. I don't think Linux will achieve "world domination" until then.

    I could be biased though. I have a strong interest in human factors. I think usability will be very important for the future computer users, the ones who dislike comptuers but will be forced to use them in their everyday life anyways. I'd like them to lose their fear and drop their attitude and learn. I'd just like to make it a little bit easier for them to do that, thats all.
    (but hey don't ever even think of removing the CLI!! CLI is good for me. It just may not be so hot for some.)

  • There's an OSS (non-free) driver out for the
    Neomagic NM2200 chipset (some Vaios, Thinkpads,
    and I think Dell's laptops) that works pretty

    (This is specifically for laptops with the
    Neomagic 256av soundcard, but without the
    optional hardware soundblaster emulation)


  • When I do get sound working the signal level is remarkably anemic. And no, Booting Linux from DOS to get the sound driver loaded is just not an option.

    Common Problem with allmost all modern laptops other than Compaq. They use the Crystal Sound stuff which does not have a convention which mixer line is which. As a result the kernel driver does not assign these lines.

    What actually has to be done is someone with more time to sit down and add them as params passed to the module at init (I know what to do but I have no time whatsoever).

  • by arivanov ( 12034 )
    The reason is simple. IBM has very large support revenue. So bringing all systems under a single OS will drop the expenses and the revenue supposedly will stay the same. Which means profit.

    Their balance sheet does not say how much did they get out of the support business last year but my wild guess is _A_LOT_... And deploying linux they will make that _A_LOT_ to be _EVEN_MORE_

    They already tried to do this with OS2 but failed because they could not port it to all architectures. Now they are trying to do this with linux.

  • Last I checked, Notes and Netscape weren't the same company. So netscape doesn't count as Lotus Notes Client.

    Bad Mojo
  • You were doing good till you forgot this was IBM and IBM has ... (dramatic pause) ... LAWYERS! Not to mention that IBM likes to manipulate it's end users by locking them into proprietary solutions. Such as the RS6000 product line.

    Bad Mojo
  • That's domino server. There is no Linux Notes Client.

    Bad Mojo
  • Linux more modern than NT ? No way ... Nt is much better designed ( at least in theory .. in practice they have still tons of issues to iron out before this thing becomes as stable as your average Unix)
  • Please IBM? SUPPORT linux on ThinkPads. Please?
  • Not that I can help you with anything else (I'm a Windows user who knows *about* Linux, but hasn't had the time/inclination to actually try setting it up), but yes, there is a USB port. It's under the fan on the left side, all the way to the back of the laptop.
  • While I can't find the link right now... I was under the impression that Linux was supported on the 600E other than the modem. Is that your only gripe, or is there something else?
  • My understanding is that Linux doesn't currently run under OS/390, but rather it is able to run in it's own LPAR on S/390 or as a guest under VM. If I'm wrong please correct me. I'd love it if IBM makes Unix System Services under OS/390 compatible with Linux. That and some way to have USS be an ASCII character set would be great.
  • Linux has won broader mainstream acceptance during the past year as an alternative to Microsoft Corp.'s dominant Windows software, especially for running the latest Internet business tasks
    So Linux is now mature for the latest business tasks!

    Linux is a modern version of the Unix operating system
    We agree (maybe Minix would have been more modern). BTW: who said NT is new technology? Linux is more modern than NT (I always told ya!), while NT still isn't multiuser. Will Win2000 be a multiuser OS?!?

    The next generation of e-business will see customers increasingly demand open standards for interoperability across disparate platforms and IBM's growing embrace of "open systems" -- software whose features are designed by a growing body of independent programmers is part of a multi-year undertaking to make IBM computers more adaptable to the changing industry
    So Open source, open standards and open systems are the way to go. We already knew that. :-)

    Well, it's not really news. IBM is committed to Linux for quite some time now, and in a few weeks its flagship software product for e-commerce, Net.Commerce 4.0 will be avaliable for Linux too (I'm waiting for it!).


  • Well, that's off-topic, but:

    If you research prior to posting, you will never become a "first poster" and your contribution will fall far behind and be read only by a few moderators, so your post also won't be moderated up among the first.


  • Sounds exactly like what CmdrTaco told me when I asked him when he was going to release Slash sourcecode. I'd sent an idea for a poll:

    Why hasn't Rob released Slash:
    1. He's afraid of loopholes in SLASH that will be instantly exploited
    2. He knows that the code is crappy as hell, and is moderately ashamed
    3. He's in the Cathedral, slooooowly planning to move to the bazaar (too slow?)
    4. doesn't want him to release code for fear of being cloned
    5. the density of adult language comments is surprisingly high
    6. "He's a hypocrite!" shouts /. readership

    To the last he replied:
    "My personal favorite. Of course, as always, every email bitching about the
    Slash code release delays the release by another 24 hours."



  • I have been running Debian GNU/Linux on my TP-770E (9548 series) for over 6 months now -- I just love it. Granted, there were some pains with it at the beginning, disappointment and frustration... But now it just rocks on!

    I'm fortunate enough not to have a freaking mwave modem (and that bastard by far not the easiest thing to properly configure and setup under NT either), an ordinary PC-card modem works great. There are things in the wish list though:

    • accelerated server for X -- XF86_SVGA, even with mode_accel still sucks
    • suspend is freaky: never works from within X, shaky otherwise
    • (this one, I believe, goes for all notebook users) better PC-card support, especially ethernet

    A word of caution: NEVER use IBM EtherJet card with it. Better never use it at all. Had all sorts of problem with it: a driver for it is not a part of standard pcmcia-cs package, it is slow -- Samba and NFS just choke on it. I use 3COM now, and am a happy man :)

  • Absolutely. I couldn't agree more. We're theoretically supposed to be able to support Linux where we are (if a customer asks), but we still have many people in our area who think it is something which runs under Windows!
  • I think you're missing a point here. IBM obviously is much more interested in servers than it is in desktops. So why bother?

    Even more: I wouldn't be surprised to see IBM quitting the desktop market altogether. They've already closed their stores, and moved to web sales exclusively. And they're not making (a lot of) profit on them anymore, quickly losing marketshare to other companies.

  • These announcements are generally meaningless. Another example: Dell. Like IBM, they tout their Linux commitment. However, their hardware plans were set long ago, and those plans were not made with Linux in mind. Like IBM's Thinkpad, which has a modem unusable under Linux, Dell has just released a server (the 2400) which has a RAID controller that has no Linux driver. Go to and try to buy one bundled with Linux. You can't, even though Dell's press releases state that all of their hardware works with Linux.
  • I'm just tired of laptops right now... The University I'm going to attend ( will require me to lease (yes, that's lease) a Thinkpad from them for only $380 per semister. Or just $760 per year, and just $3040 for four years! Then I give it back!! If I want to keep it--that's another $1600!!! We're up to $4640 now just for a Thinkpad 1412i...

    I can't turn it down, can't use my own, can't say I'd never use it. I can't even get rid of Windows!! What will I do?!

    Has anyone else experienced anything like this? What have you done? Are there any ways out of this?
  • Last time I spoke with my Netfinity sales rep, they told me that the Netfinity group was part of the Enterprise Servers group you mentioned. In fact, they bragged that that fact was enabling the process of integrating high-availability features like Chipkill [] into the Netfinity servers.

    Am I supposed to be impressed that they're not?
  • Buying binary drivers are all nice in theory, but guess what? Every new major kernel release (and some minor releases) will _REQUIRE_ that you buy new drivers to work with that kernel.

    Yep, you read me right: portable binary kernel mods will _NOT_ be supported by Linus (gleaned from his statements on l-k), and therefore by linux. Drivers are, by necessity and design, kernel mods. Therefore, since the kernel module APIs are going to change between version, binary module/driver compatability will not be guaranteed at all.
    Basically: if you have a piece of hardware and want to provide a driver for it, be prepared to either have a programmer or two always ready to forward port it, or be prepared to OS it.
  • There's no reason why CMVC shouldn't exist on Linux. I believe there's already a CMVC client on Linux although I haven't tried it yet.
  • NT4 Terminal Server Edition is a multiuser environment, and two of the three editions of Win2000 have Terminal Server built in. So I guess the answer is Yes.


    "You can't shake the Devil's hand and say you're only kidding."

  • Who cares if it's moderated up!?! Do you really need the Karma that bad? Are you so starved for attention that you need *everyone* to see you post re-parented?? Jeeeeezus!


    "You can't shake the Devil's hand and say you're only kidding."

  • Yes you are right, but I'm stuck with using AIX 4.1 since that is what my customers have :(

    And I know better than to tell a customer what to do, even if it is for the better.

    Steven Rostedt
  • I liked this comment from the article...

    "Linux is a modern version of the Unix operating system, the software widely used to control the powerful computers that manage central business operations at many companies, but its suitability for running desktop computers remains in doubt. Still, Linux has won broader mainstream acceptance during the past year as an alternative to Microsoft's dominant Windows software, especially for running the latest Internet business tasks."

  • Dear IBM,

    Please, please create a usable open source java environment for linux.

  • Are they going to contribute to the community, or are they just going to make a quick buck on everyone else's work without having to worry about NT licencing fee

    I'm not sure we want to have IBM too heavily involved in the development process... their methdology is nothing like what has built the kernel so far.

    Rather than ask them to develop Linux, which will only upset them (and upset other jagoffs like Michael Dell), let's ask them to really open up availability of hardware specs and info. Then the kernel hackers can do what they do best, on the best hardware, and Linux will be further entrenched.

    Just my Christmas wish eleven months early... Please, Big Blue, use Debian, not Red Hat!

  • I can understand your point, but I think the
    Alpha/NT example is a little off base. If I'm
    going to drop $80k on a machine, what's another
    $3k for an OS :) Besides, the Q doesn't recommend
    Linux on the 8400, they recommend Tru64 or VMS. And that's not free (beer or speech).

    This is more of an issue for price sensitive markets like home users.

    P.S. I don't own and 8400, so I'm only stating what I recall it to cost. And I've never gotten a price quote for NT Enterprise Edition, so I've
    got no clue what that goes for either.
  • Umm I don't know where you got your information but right there on the IBM site it tells you there is an Enterprise edition of WebSphere for Linux.
    >Build, run and manage high-volume Web sites with
    >WebSphere Application Server Enterprise

    Try reading a little closer before you try and tell the world what is plainly written on the IBM web site.
  • Yes, OS/2 is doing very well. My company has been deploying it in the banking industry for some time as a server and desktop(much stabler than M$). And wouldn't most of you be supprised to find out you were using OS/2 every time you use that nifty ATM to get beer money!*grin*

  • by Anonymous Coward
    Look at and Hardware drivers are 90% ready. We need someone to write v.34, linmodem project is not moving too far. Pavel Machek --
  • I think most big IBM customers don't like doing really rapid changes in systems (eg. dropping SNA and scrambling to get a TCP/IP version of their system in place). I think it would be a good thing for IBM's customers if they would open up SNA, perhaps offer source for an SNA driver for the Linux kernel. That would allow for bridges between SNA and TCP/IP systems, and would give a nice migration path for people who want to move away from SNA in a slow, well planned fashion, as well as better functionality for people who are fine with SNA and want to stay there.

  • Expect to see price hikes in other IBM Operating Systems as these get smaller market shares. Who knows, they might even Open Source some of the marginal Operating Systems, like OS/2.

    Unless I'm mistaken, IBM can't Open Source OS/2 because Microsoft still have a firm claim on much of the code. A lot of folks have been trying to get them to release the WPS code, but for all I know, it's subject to the same problems as the OS.

    What will be interesting to watch is how IBM treat AIX during the next year. If they begin porting all the enterprise goodies to Linux, I expect we'll see the other UNIXs climbing over each other to get their stuff ported. This could be enough to convince Sun to come clean and get serious about Open Source.

    Wouldn't you like to be a fly on the wall in Microsoft's board room today?

  • Thing is -- there's a number of websites out there which started off on an old notebook. The idea of what is and isn't a server is really doing a headspin right now. My last consulting gig relied on a "server" that would be eclipsed by most of last year's notebooks -- a dual 90MHz HPUX box with 256MB RAM and about 8GB storage. This was a statistical analysis server for about a half-dozen statisticians and programmers at a major pharmaceutical company (granted, it was a gimpy old box the department had been saddled with).

    And from all I've seen, Thinkpads make really spiffy Linux boxen, if you're willing to accept a few warts. It would be very cool if those warts could be removed.

    And yes, I agree that the issue is more one of IBM being a large number of fairly autonomous divisions. Still, a unified vision in this one area would be most sweet.

    What part of "Gestalt" don't you understand?

  • Jeff Papows resigned from head of the Lotus division in December. His behaviour was so strongly pro-Microsoft that conspiracy theorists thought he must be in their pay. Now he's left, we might actually get somewhere.
  • IBM just contributed patches to the kernel to run on the S/390 hardware. That counts for something, doesn't it?

    See what they've done before berating them saying they're "just going to make a quick buck".
  • Just remember - as far as the Dinosaur Folks are concerned, they've been around forever and it's up to us *IX types to work with them. Not the other way around. In that light, it's absolutely fascinating that the Dinosaur Maker itself has put such a wide-spread stamp of approval over Linux...

    Some of the big IBM decision makers were techies back in the early days of mainframes. My boss was recently telling me how in the early days, the source code to mainframe operating systems was freely given to customers, and customers would frequently provide a fix along with any bug reports. Then gradually, it became de-rigeur to keep your source code in a locked box.

    In light of that, you can see how Open Source just might click with that generation.
  • Admittedly, most of the programming world is far more familiar with TCP/IP. But this is not an argument for getting rid of SNA. SNA was developed and optimized for the mainframe architecture, and there is no doubt that it does what it needs to do extremely well.

    The argument for knocking off of SNA would be if TCP/IP is shown to have superior performance characteristics than SNA - something which is not the case. The protocol is indeed proprietary (which is bad), but it is not screwed up. You seemed to have made that statement solely because you, personally, are unfamiliar with SNA.

    And as another poster has pointed out, TCP/IP on IBM mainframes does exist, and has existed for quite some time.

  • My IBM Thinkpad 380Z Works fine running Mandrake Linux 6.0 and Caldera OpenLinux. I've got an Viking 56k PCMCIA Modem and an Linksys 10/100 Ethernet card in it, and both of those work great too. The only thing that doesn't work on mine is the 16 bit sound. The 8 bit sound works fine. The only thing I had a problem with because of that was playing all my MP3s.
  • Bravo! Someone moderate this up to its full 5. =)
  • IP QoS and more particularly CoS (class of service, most standard version being DiffServ) is coming - it's just requires policy-based management tools to make it much easier to set up.

    SNA over (say) the DLSW protocol over IP with CoS is quite a good combination - you give low latency to the important traffic, particularly the network control stuff, using DiffServ priorities (using the TOS byte in the IP header).

    However, most IP CoS products don't let you prioritise between different SNA traffic - e.g. printing and transactions get the same priority - so it's not quite there yet, though Cisco routers at least do let you do this.
  • Yes, OS/400 has TCP/IP; and it's still buggy, and about two years ago they finally got it to handle 27 whole Mbps over Fast Ethernet.

    Their TCP/IP service daemons often suffer from brainlock, in far greater frequency than any decent Unix including Linux.

    People don't rewrite their legacy apps to use TCP/IP because SNA is more reliable on OS/400.

    I can't speak to OS/390.
  • But what are they going to do to develop linux?

    The list is long. Very long. Do you know about Jikes [], for starters? how about Linux on the 390 []? It goes on and on.
  • This isn't IBMs problem. Customers still demand SNA. In fact, it's considered a huge change in the mainframe industry that versions of MVS that don't have TCP/IP are considered obsolete and unsupported by IBM.

    SNA also has a few advantages over TCP/IP.

    With TCP/IP, you have to overprovision your network by a fairly large margin in order to handle peak loads. The way SNA works, those peaks don't happen, and so you don't have to overprovision as heavily.

    Also, for similar reasons, response times are more predictable with SNA than TCP/IP, so it's more suited to certain kinds of real-time response applications, such as airline ticket sales. :-)

    Not that I'm a big booster for SNA. It's a stupidly designed protocol that deserves a quick death, but it isn't going to happen anytime soon.

  • OS/390 aka MVS has had a full POSIX personality for ..err. another long time (don't know how long). This was specifically put in to enable "ease of migration". When you see an Apache Web Server running on S/390, it's using the POSIX environment. You'll note that all of IBM's middleware (the DB2 and MQSeries-es of this world) have also been TCP/IP enabled since Pontious was a Pilot.

    So from a "killer app" for migration SNA -> IP, all the components have been in place for a number of years. That there hasn't been a mass exodus from SNA-based applications indicates to me that one or more of the following conditions therefore apply:

    • There's no money for migrating legacy applications to IP
    • There's no point migrating perfectly acceptable and working SNA apps to IP
    • SNA works well in it's environment and either has no significant deficits to or indeed possesses positive advantages over IP when dealing with Mainframe applications.

    With regards to "skills gap", I'll probably cause a flame war here by calling you an upstart UNIX weenie here.. Remember that S/390 has been "out in the wild" running large-scale commercial installations for 30 odd years. There's a *lot* of skilled people out there who can do COBOL, who can administer IBM mainframes, and who get paid good money for putting in and maintaining SNA networks.

    Just remember - as far as the Dinosaur Folks are concerned, they've been around forever and it's up to us *IX types to work with them. Not the other way around. In that light, it's absolutely fascinating that the Dinosaur Maker itself has put such a wide-spread stamp of approval over Linux...

  • Ah, OK I think I mis-understood where you were coming from then. I apologise.

    Supplementary question for you though: You're big on moving to IP to fix networking problems. Is this because:

    A) TCP/IP is inherently more stable, scalable and better at traffic management than SNA(*)

    B) TCP/IP networks are "cooler", and much more importantly more widespread, than SNA networks and therefore the thrust of both market place development AND corporate IT strategy is heavily favoured towards IP instead of SNA?

    Obviously the implications are the same in either case - move to IP - but I am genuinely interested in getting an answer to this question from someone who really understands both sides.

    (*) = My understanding based on what I've heard is that SNA is much better at traffic control and prioritisation than vanilla TCP/IP.

  • Well, I can't be authoratitive on this, but I know that at some point in recent history, Lotus' own strategy was that the Notes client would disappear to be replaced with a Web interface to all Notes functions. Indeed, you CAN get your email, and browse databases etc, across the web - if the server is enabled for such. You can also get POP3 access to your mail if enabled. You may want to check with whoever runs your servers to see if that's enabled.

    Mind you, then Lotus spoilt it all by saying "Web access everywhere. But, er.. Well, we'll do a Windows client for Notes 5". Since that covers probably 99% of the target client audience, y'all out there running "non standard platforms" on internal systems with the web access goodies turned off are all a bit stuck...

  • Slim: the situation you describe wasn't quite Open Source... Source code was never "freely given" to customers, although it was a core part of the deliverables that customers received when they licenced software.

    The uproar caused in the User Groups when IBM finally switched to binary-only licences in the early '70s was an unpleasant sight to behold, apparently....

    I really couldn't hazard a guess at how the IBM management view the Open Source movement. Call me an old septic but I rather suspect they're motivated more by visions of breaking OS strangleholds and increasing marketshare and lucrative services opportunities than they are by any ideological considerations...

  • D) is true, except for the fact that the 3rd party SNA stack my Windows stinkpad is currently running is from... IBM. Developers, maintainers and specifiers of the SNA protocol. And, incidently, developers & maintainers of the SNA stack on the S/390 systems to which I'm talking.

    So, yes it's 3rd party with respect to the O/S manufacturer, but then SNA is a 3rd party protocol to my O/S manufacturer. And I've already seen how good they are at implementing other not-invented-here protocol stacks. Like TCP/IP.....

    Aside from the above commentry, your implied conclusion - SNA stacks on desktop machines aren't stress-tested - is irrelevant to their intended function. If it connects to a Mainframe, it's performing probably 100% of it's intended function(*). If it does so reliably, under variable simulated or tested networking conditions, then it is Fit For the Purpose It Was Bought For. Comparisons to the tests vis-a-vis IP stacks and relative number of users don't factor in.

    (*) = unless you're running one of IBM's early-to-mid '90s desktop-OS APPC applications like NVDM/2 or DB2/2 that is. In which case, getting a 3270 connection to the S/390 is just the *start* of your trouble, and it's time to now start editing, then compiling, then activating, then testing innumerable options in a text APPC configuration file, begging your SNA networking guys for the appropriate magic numbers, and generally cursing your miserable existance on the planet. Not that I have personal experience of this of course.....

  • Someone already put in a C so I'll start with...

    D) A TCP/IP stack on a client desktop machine is generally a more mature solution than a SNA stack on a client desktop machine.

    If you're connecting to a machine from a Windows box via SNA, you're using some kind of 3rd party stack. It gets used, primarily, to connect to mainframes. And not much else. So it doesn't get tested much, at least not on the same scale as TCP/IP gets tested every day.

  • These announcements are generally meaningless.

    they mean a lot to Wall Street, which I would say is the sum total cause for this "announcement" (action to be done at a later date)
  • Fisrt, I'll set the stage by saying you won't find a bigger IP bigot than I am - I've been doing IP/Internet stuff daily since 1985, did the IP migration for a major oil company, and have been an Internet consultant. (Maybe more appropriately I should say I'm a bigot for the whole inet/Unix philosophy of keeping things simple by pushing the intelligence of the net to the periphery.)

    That said, I developed a deep respect for the mainframe guys along about 1992, when we were hacking the mainframes into the IP intranet. (And anyone that's ever worked with an IBM 3172 protocol processor (because IP protocol processing is death to a batch-tuned machine) knows what a hack it was, especially early on.)

    Although their methods and thought processes are quite different from the Internet "norm" (wow, I'm mainstream now - dangerous!), they came up with very good, very valid engineering solutions to the problems they faced at the time. SNA networking is a royal pain, and I avoid it where ever I can (which is pretty much everywhere by now). You try building a network which must be bank and hospital reliable, and connect thousands of users all over the country, often with only 9.6 kbps lines to serve an entire site!

    SNA is elegant in its own twisted way, and as a network guy, I can certainly appreciate why it worked well then, and why it may well be with us for quite some time. Granted it's a pain to work with, but IBM and the other mainframe folks have very good IP support today, so it's really a non-issue. What difference does it make to you if the back-end network between the mainframes and their storage/output devices is SNA? It's (fairly ) simple, incredibly robust, and dictatorially controllable - even the IP bigot I am, I have to confess I envy the SNA guys' ability to manage and control traffic with fine-grained and even adaptable policy.

    The moral is this: Don't be too quick to shoot down competing technologies until you learn a bit more about why they're there. (I say this because I could have been reading my own rant about SNA circa 1991 in one of the earlier posts...) We can learn a lot from the mainframe guys, and vice versa - and that's what makes what's happening now so much fun...

    (Disclaimer: I work for Tivoli, an IBM company which does all kinds of cool systems and network management stuff for pretty much everything. We get free beer on Fridays, too... it's not your daddy's IBM.)
  • It seems from the article, that Mr. Gerstner thinks Linux is a Java, and that programs that run on Linux can run on all Linux's everywhere. Which of course is false.

    OK, that is what it looks like from a first glance view of the article, but I know better. IBM is trying to go back to their old philosophy of selling hardware. Let someone else do the software. IBM is putting all the money in so that it will take off. This is a good strategy, since it could take the dominant force away from their foe "MS". As well they look like the heroes, and their old reputation will finally be gone. Thus, they can claim it back again!

    I'm worried about VA Linux. Don't get fooled by the stock price, they can be crushed by IBM. RedHat probably has nothing to worry about this, infact, they may make out more by this.

    Don't get me wrong, I think this is a Good Thing(TM). Maybe now I won't have such trouble compiling GTK on AIX!

    Steven Rostedt
  • Does this mean I might be able to resurrect my CISC AS/400 model 9404? Even though it's obsolete, IBM still wants thousands of dollars for the O/S for this beast.

    There are tons of old CISC AS/400's out there that could be put to good use if there was something other than OS/400 (for the price) to run on it. Since the "Linux on AS/400" project seems to be going nowhere, this might be my only hope.

    Even if it never runs again, my 9404 makes a K-Rad night stand.

  • Why would anyone want to support their legacy
    apps when they could throw away millions of dollars to run Linux and be kewl? :)

    I think your point about legacy support is a little over the heads of many /. posters.

    Since this will be moderated down to flamebait, perhaps I should mention hot grits and Natalie Portman :)
  • This is really out of IBM's hands, If they dropped SNA their current clients would revolt. In some Enterprise environments the legacy equipment, ie green screens, printers, which work perfectly fine for the application needed, will only use SNA. Anyone ever hear of a dumb term with an ip address? SNA is a logical, efficient protocol that coexists rather nicely with TCP/IP. DLSW is a prime example of the best of both worlds. SNA has made my brain sweat on more than one occasion but no more than any other protocol. In fact I think IPX takes the brain fart award and would much rather see it go away. At least SNA makes sense.
  • If you look at IBM's product lines, their agenda is fairly evident. WebSphere for example.

    They have three levels Base, Advanced and Enterprise.

    Base = Apache + a few bells and whistles. (Linux, Sun, AIX and NT)
    Advanced = Base + EJB + Java ORB + DB bells and whistles (no Linux alone)
    Enterprise = Advanced + CB (again no Linux alone).

    They want to keep the fingers on all the pies (nothing wrong with that). But if you need to scale, you need AIX - for which you pay an arm and a leg - of course you get corporate support and the developer base from Linux/Unix.

    This is the same strategy IBM would adopt across all product lines. If you see a product from IBM that has "universal feature availability" on all platforms (DB2 for eg - go on correct me) then they either have too much competition in that area or it doesnt have a value a proposition for AIX.

    Just look at their Java 1.2 release strategy
    Linux version in 2nd qt. '00
    NT version 1st qt. '00
    Aix version last qt. '99

    Bottom line AIX and OS/390 are excellent revenue streams. They will never hurt them.

    -- The best of my sons ? oh the one setting alight the roof top.
  • BBC News [] also has the story [].
  • I'm curious what they will port to Linux from their other platforms. Even if they keep their latest stuff proprietary, IBM has quite a few older tools that have been pretty good. Regardless of whether you love or hate the interfaces, protocols and data formats, they have tended to be very robust. It would be nice to see some of the older ones appear as open source. For example, while I love CVS, I could imagine a couple of the features of CMVC being added to it.
    • IBM can't Open Source OS/2 because Microsoft still have a firm claim on much of the code.

    I'd forgotten this. Yep, don't expect an OSS OS/2 soon. At least not until Microsoft starts releasing Open Source.

    • Wouldn't you like to be a fly on the wall in Microsoft's board room today?

    I doubt if there's anything in this that raises many eyebrows in Redmond.

    IBM has, months ago, heavily committed to Linux in the areas that compete most directly with MS offerings. This is about commitment to Linux in the rest of the IBM line.

    IBM also made quite a splashy commitment to Windows 2000 last year.

    I think this is more about IBM deemphasizing their own OS line vs. taking up Linux against MS.

    IBM may be banking on a shakeout that leaves only Linux and MS standing. If such a shakeout occurs, it's the various UNIX and other OS vendors that'll need to scramble.

    -Jordan Henderson

  • The argument for knocking off SNA would be that you have to contort your network in all sorts of non-optimal ways to deal with the fact that you have, essentially, two networks ... one for SNA traffic, and one for the TCP/IP apps that the rest of your business uses, yet you need to provide connectivity between nodes running both stacks.

    It doesn't seem to be useful to me to be running your POS systems, or reservation systems, etc. on SNA just because that is the way the original app was written. Using classic SNA, you are talking about a connection-oriented protocol ... which is clunky, complex, and expensive. The idea of a mainframe-controlled connection may have been useful when LANs were running at 4Mbps, but the resources available now invalidate the "need" for predictable response times provided by SNA. So IBM comes up with APPN ... not a lot better, even though it will do route discovery and allow you to merge traditional LAN traffic with your SNA mainframe sessions across the backbone, but it is once again clunky, complex, and not very efficient.

    A far better fix would be to use QoS/CoS across an IP backbone and allow the mainframe (and associated apps) to deal with it's sessions over IP ... thus freeing everyone from the connection-oriented crap and allowing the mainframe to play on a LAN as if it were another server.

  • "With regards to "skills gap", I'll probably cause a flame war here by calling you an upstart UNIX weenie here.. Remember that S/390 has been "out in the wild" running large-scale commercial installations for 30 odd years. There's a *lot* of skilled people out there who can do COBOL, who can administer IBM mainframes, and who get paid good money for putting in and maintaining SNA networks."

    No flame war there, as I don't claim to do any sort of programming or operation of IBM mainframes ... I do, however, see networking environments that support SNA as their method of connecting to said mainframes. I don't much care if the apps are still running strong, as the networks that support those connections are falling apart under the stress of corporate LAN requirements. If something provides me some shred of hope that people will upgrade/migrate, and move to an IP based network, I cheer it's arrival.

  • No apology necessary ... I'm pretty thick skinned. As to the supplementary question (and this comes with some caveats):

    My answer is C) TCP/IP WITH the addition of QoS/CoS across the backbone allows for a more effecient backbone design, and removes the need for multiprotocol networks which support both IBM protocols (SNA, APPN, all that junk) and native IP. The primary reason we would want this is B) that IP networks are more widespread, and increasingly required to interoperate with traditional SNA networks ... and said interoperability is clunky, a pain in the ass, and generally slow and suboptimal.

    * SNA is much better at traffic control than TCP/IP without quality of service. However, the need for that level of control should have been erased when the rest of the world migrated to switched Fast Ethernet (god only knows when that became available to FEPs ...) and QoS/CoS can do a very good job of providing predictable response times, guaranteed bandwith, etc.

  • The reliability to which you refer is the main reason large companies went with SNA in the first place. Happily, I have seen results that are at least as good using IP QoS/CoS to provide consistent response time and bandwidth.

    The QoS/CoS capability (at least on a Cisco platform) is the driving force for Voice over IP and Voice over Frame Relay that many businesses are rolling out. If I can engineer a network that will provide predictable bandwidth and latency that is good enough for a human ear (less than 200ms latency, 13kbps dedicated bw per call), then there should be no issue providing consistent response time to a TN3270.

    I have to agree, though, that it probably won't go anywhere soon. The tools have been in place for a long time to phase out SNA, and it hasn't happened ... I attribute this mostly to old mainframe operators who refuse to learn a new technology and maintain a 30-year old death grip on the server/network environment.

  • I wouldn't necessarily call DLSw (or even DLSw+) the best of both worlds ... yeah, it provides some interoperability, but not with fast or well.

  • Gak! typo. Read "not fast or well."

  • You are entirely correct that there is TCP/IP support. I believe an open, LINUX OS on the mainframe would make software portability easier, and would make the argument to migrate both the software and protocol stack to a pure IP environment much more palatable from a business perspective.

    What I am looking for here is the "killer app" that would cause a migration, and LINUX as the OS could do that ... smooth interoperability with other 'nix servers, less of a skillset gap between admins, a single network architecture, no more "split networks" which are designed to deal with SNA and IP traffic separately, etc.

  • The other day I was visiting a fellow contractor in another part of the building and noticed he was running E and Gnome on a big screen. I asked him if it was Linux and he told me it was AIX. Apparently he'd managed to get it all to compile. The latest AIX seems to have the X11R6 support necessary to compile those neat Linux tools.
  • IBM's got a few choices for support offerings. They could go with their IGS if they can scare up enough Linux talent. I think it's more likely that they'll outsource the whole job to Linuxcare or Redhat or one of the other Linux support companies. Those companies are in a much better position to give good Linux support, and IBM's not stupid.
  • They are working on getting Lotus Notes and domino working on Linux (both server and client)and the Viavoice teams are committed to Linux and have released a speech SDK for linux ( here []

  • The Internet and electronic commerce are two of today's most important technologies. Using information technology to connect the entire enterprise has become a reality in corporations around the world, with dramatic improvements in company-wide efficiency, profitability,and accountability coming as a direct result. Today, IBM's Web site has more than 270,000 pages offering 14,000 products, is localized in 70 countries and 16 languages, and receives 7.8 million page visits per week worldwide. When IBM launched the redesign in February 1999, traffic shot up by 120 percent, and sales increased by 400 percent. IBM projects that e-commerce revenue will swell to $10-15 billion in 1999. IBM expects to procure $12 billion in goods and services over the Web, saving $240 million by replacing five million paper invoices. By supporting millions of self-service Internet transactions,IBM also anticipates saving approximately $600 million in 1999 in call center and field specialist support costs. When used to its full potential, the Internet offers businesses new opportunities to increase revenue and gain a lasting competitive advantage. VA Linux (Nasdaq:LINUX)on Dec 9, launched its $132 million IPO, pricing 4.4 million shares at $30 each. By the end of the day,those shares were worth $239 each. The company's stock gained 700 % in its first day of trading and VA Linux had achieved a market capitalization of $9.5 billion. VA Linux in Sunnyvale, Calif., manufactures machines based on processors from Intel Corp. that runs the Linux operating system. An alternative to to Microsoft's Windows NT. Jimmy Castro Member Greater Austin Chamber of Commerce Austin Texas
  • by Anonymous Coward on Monday January 10, 2000 @12:48PM (#1387387)
    Umm...not sure I agree here (and I work for IBM as well).

    To be honest, nothing at IBM involves Linux. Everything IBM has done with Linux has been essentially an external posturing and hype

    Let's see what I can rattle off.
    1. DB/2 [] for Linux.
    2. MQSeries and ADSM clients for Linux
    3. 24x7 support [] for Linux on Netfinity Servers [] through the IBM Helpcenters
    4. GPL'd device driver for out ServeRAID PCI RAID Adapters [] (and onboard versions on Netfinitys)
    5. Domino Server
    6. A fast JDK [] for Linux
    7. Jikes [] Java debugger
    8. Code released to get it to run on an S/390
    9. Websphere for Linux []
    10. VisualAge for Java now runs on Linux []
    11. IBM HTTP Server [] for Linux (part of Websphere)
    I could probably go on if I tried. Nothing at IBM involves Linux?

    As to Lotus Notes...sadly (or blissfully, depending on your opinion of the product), I don't see a Notes client happening. And while we don't do this at IBM, there is nothing stopping a customer from enabling browser- and SMTP-access to Notes databases (including mail).

    Gerstner's even gone so far as to put one of his golden boys, Irving Wladawsky-Berger, on Linux (NYTimes story here [], requires free login). Wladawsky-Berger has been credited with a lot of what got our Internet business going. For Gerstner to move him to Linux work is, IMO, a big deal.
  • by IQ ( 14453 ) on Monday January 10, 2000 @04:56AM (#1387388)
    I am running Linux-Mandrake 6.1 on my ThinkPad 600. It is just that there is no usable sound. I bought a Thinkpad without the DSP based proprietary modem that all the other thinkpads are crippled with.

    APM - forget it. - No wait, you can get APM working but you have to leap tall buildings with a single bound to do it.

    I have not tried the iRDA or USB ports (is there a usb port on it?).

    Hot swap floppy? Anyone? Anyone? I did not have any success with that. I have been running Linux on notebooks since '95 and am used to jumping through hoops to do it (debian with floppies on a gateway nomad amd 486!).

    When I do get sound working the signal level is remarkably anemic. And no, Booting Linux from DOS to get the sound driver loaded is just not an option.
  • by Geek Boy ( 15178 ) on Monday January 10, 2000 @05:16AM (#1387389)
    Virtually everything works on the new thinkpads already....... There is a Lucent WinModem driver now (which was deemed not important enough for a slashdot story when I submitted it), there is a sound driver for the new i series, the video works (although the accelerated driver is still under development and the developers at XF86 have been ignoring my emails). The DVD is usable but movies don't play in linux and that's no fault of IBM's. In any case, IBM is being VERY helpful in getting Linux up and running on Thinkpads. Mine runs great!
  • by warpeightbot ( 19472 ) on Monday January 10, 2000 @06:39AM (#1387390) Homepage
    I think it would be a good thing for IBM's customers if they would open up SNA, perhaps offer source for an SNA driver for the Linux kernel. That would allow for bridges between SNA and TCP/IP systems, and would give a nice migration path for people who want to move away from SNA in a slow, well planned fashion, as well as better functionality for people who are fine with SNA and want to stay there.
    Already been done, by Turbo Linux. Token Ring and Ethernet drivers, TN3270/TN5250 servers, tech support, the whole nine yards. All GPL'ed. Check it out on [].

  • by henley ( 29988 ) on Monday January 10, 2000 @04:43AM (#1387391) Homepage

    Ummm.... OS/390 and OS/400 have had TCP/IP for a *long* time.... I was using a VM system in 1993 that had IP.

    Perhaps you mean "why can't people re-write their legacy apps to use TCP/IP instead of SNA?". Which expands the scope of your complaint to encompass more than just IBM I think...

  • by JordanH ( 75307 ) on Monday January 10, 2000 @04:55AM (#1387392) Homepage Journal
    I don't understand your point.

    IBM has claimed to support Linux on their laptops for some time now. And, they've supported Linux on the PC Servers for a awhile also.

    This announcement is about Linux EVERYWHERE, going UP their product line. We've heard about Linux on the Mainframes. Now, I guess this means AS/400s will get it too. And, it's a firm commitment on IBM's part to support it on RS/6000 machines.

    I think IBM believes this will return their economics back to the mid 60's, when they dominated the market and gave away their Operating Systems when you purchased their hardware. The problem with that was that the DOJ found it to be anti-competitive and forced them to start selling the OS's unbundled from the hardware at "competitive" prices.

    Nobody since then, with the exception of MS, has ever made much money selling Operating Systems. It's capital intensive and your market for new releases can dry up really fast.

    I'm sure IBM realizes that they can't hope to make money in Operating Systems the MS way, so... IBM is deemphasizing the Operating System as a profit center. Let others (MS and the Free Software Community) deal with the headaches. IBM will still invest heavily in making sure the dominant OSs run well on their hardware, but they can greatly simplify their lives by letting others develop the "commodity" Operating Systems.

    I think IBM also believes that their services business will pick up in supporting end users and Enterprises on Linux. IBM is quietly but aggressively pushing into Linux services. This services business is very lucrative.

    Expect to see price hikes in other IBM Operating Systems as these get smaller market shares. Who knows, they might even Open Source some of the marginal Operating Systems, like OS/2.

    -Jordan Henderson

  • by netpuppy ( 77874 ) on Monday January 10, 2000 @04:34AM (#1387393) Homepage
    If they are serious about the mainframe part, and about open standards, maybe it will be a good excuse to migrate all of those poor bastard banks, retail chains, and airline reservation systems that still run on (shudder) SNA. Everyone else runs TCP/IP ... why not IBM mainframes?

    Imagine a world where us poor network engineers don't have to cope with screwed-up proprietary IBM network protocols in the data center. (starry-eyed sigh)

  • by Greyfox ( 87712 ) on Monday January 10, 2000 @05:10AM (#1387394) Homepage Journal
    A lot of the Linux stuff seems to be driven by single employees. Linux infects the engineers first. How these announcements get made is that the engineers ally with marketing(!) and bypass the lower managers who fear innovation and are too myopic to see the big picture.

    The key to manipulating marketing people for your own evil goals is to drop a few buzzwords, "You know, more of our customers could use this driver if it were Open Source!" (Something you tried to slide by your manager and got shot down for) and marketing goes "Ooo! Open Source" in that typical Dilbert marketing tone. The other marketing people take up the battle cry and the next thing your manager knows, his second line is asking him why the driver isn't open source. Your manager then comes to you (having completely forgotten shooting you down the first time) and asks you (in that same irritable management tone as if it's your fault) why the driver isn't open source. You are ready for this question and say "Well I've been considering that and there's no real reason it couldn't be..." and Vola! Open source product.

  • by Bad Mojo ( 12210 ) on Monday January 10, 2000 @04:52AM (#1387395) Homepage
    I work at IBM. Everyone outside of IBM tells me, "It must be great working at IBM with all the Linux stuff they do." To be honest, nothing at IBM involves Linux. Everything IBM has done with Linux has been essentially an external posturing and hype. Otherwise you would see Notes for Linux, Lotus Smartsuite for Linux, and Linux would run properly on RS6000 machines with Token Ring.

    IBM might want everyone out there to use Linux, but until I see IBM using Linux, I won't believe it.

    Bad Mojo
  • by ryand ( 82763 ) on Monday January 10, 2000 @06:56AM (#1387396) Homepage
    I'm going to try to respond to as many of the comments as I can: IBM is supporting Linux for 1 main reason - their customers are asking for it. They've never hidden that reason, and are quite open about it. Unlike most companies, by supporting it, they mean also contributing developers to it. Believe it or not, there are some _very_ experienced developers in IBM working exlusively on contributing source to Linux, Apache and maybe some other OS projects. This is costing IBM a lot of money, so it's not to be taken lightly. It's also a long term investment, because they don't make very much on Linix based projects right now.
    "no one can use it on the Desktop until Lotus comes up with "Notes for Linux""
    I don't know why Notes doesn't have a Linux client. I use it, and I can honestly say that I felt better before I used it. My personal opinion is that I hope that someone else comes out with a Notes client for Linux, because the Notes client I use is pretty awful.
    "To be honest, nothing at IBM involves Linux. Everything IBM has done with Linux has been essentially an external posturing and hype"
    Just because nothing you do has anything to do with Linux, doesn't mean that nothing _IBM_ does has anything to do with Linux. Come on - it's a huge organization. Most of the Linux work is being done in a grassroots way, and once it becomes big enough it becomes officially endorsed by the upper level Linux initiative. I'm hoping to soon be involved in a movement of my project to Linux. It's starting off unofficially (I'll be working on it in my spare time) but hopefully, it'll be part of everything else I do, and eventually officially endorsed. Why am I doing it? because it'll be a challenge, and it'll be fun to see what I'm working on run on Linux. My point is, if you don't see anything done with Linux, do something about it yourself. Even if you haven't been asked to do so.
    "until I see IBM using Linux, I won't believe it"
    Well, I've seen IBM using Linux, and I've only been here for a few months so I believe. Look around a bit more, you'll find it soon enough.
    "don't have to cope with screwed-up proprietary IBM network protocols in the data center"
    I'm not sure about this - wasn't there a time when that protocol was better than what else was out there? Didn't it make sense then?
    "But what are they going to do to develop linux? Are they going to contribute to the community, or are they just going to make a quick buck on everyone else's work without having to worry about NT licencing fee"
    IBM's Linux site [] It might not say anything about the contributing developers, but at a conference CASCON, the person in charge of Linux projects at IBM talked about contributing back to the community at length.
    "Who knows, they might even Open Source some of the marginal Operating Systems, like OS/2"
    Just a note - OS/2 is not that marginal. Actually, it's doing well from what I hear. I don't know much more, but I doubt that it'll be open-sourced.. Disclaimer: I'm a relatively new employee for IBM, and I don't claim to speak for them at all. That's not what I'm paid for, and I might have things wrong, so it's a good thing I'm not getting paid for that.
  • by Greyfox ( 87712 ) on Monday January 10, 2000 @05:01AM (#1387397) Homepage Journal
    IBM's doing some neat stuff and they're giving a lot of it away for free. They've got a bunch of XML tools, a Java IRC program that can be embedded in a web page, their own Java VM which may or may not be more advanced than the Blackdown one (I haven't checked lately) the Postfix sendmail replacement and REXX and Object REXX for Linux.

    At least some divisions of IBM are "with it." I'm pushing to open source some of the UNIX stuff I'm doing for them and hope to start shoring up some areas where Linux has significant weaknesses if I can get the ball rolling on some of these projects.

  • by cruise ( 111380 ) on Monday January 10, 2000 @06:19AM (#1387398) Homepage

    IBM: You'll get it when we're finished and if you ask us again, we'll delay it another week.

    They are a threat to free speech and must be silenced! - Andrea Chen
  • by Silverpike ( 31189 ) on Monday January 10, 2000 @06:43AM (#1387399)
    Disclaimer: I am an IBM employee

    Um, I think my friend Mr. Malda has confused some /. readers. The announcement, as stated, applies to Enterprise Servers, which in IBM lingo means RS/6000, AS/400, and OS/3[7-9]0 machines.

    As far as I can tell, this does not affect notebooks, PCs, and Netfinitys. They fall under a separate division of IBM and have their own "master plan". This is somewhat moot however, since Linux does run fairly well on these machines anyways.

    As some readers insightfully pointed out, there are obvious motives for this. AIX, VMS, and VM are expensive to develop and time consuming to maintain, and IBM makes more money off the hardware anyways. IBM still has very strong hardware expertise, and the best reason to buy a RS6k is the hardware architecture (that and all the reliability aspects).

    Don't have the misconception that IBM's enemy is Microsoft. Although we compete with them, our real competitor is Sun. Sun competes heavily in all the same areas we do, and Linux is the perfect way to help us fight the the workstation battle.

    Since it is obvious to me that Sun has no intention of really supporting Linux until it begins to threaten their survival, I'm all for IBM and Linux partnership. This means IBM will contribute to linux kernel development for all of the products mentioned above, which should be quite valuable to Linus and Alan.

    As for applications, that too falls under a different IBM division. I can't tell you if Notes or Smartsuite are coming for sure, but I wouldn't be suprised to see some changes in light of this announcement.
  • I work in development for Big Blue and lots of us here love Unix in general and Linux in particular but no one can use it on the Desktop until Lotus comes up with "Notes for Linux". Now mind you no one actually likes Notes here, but if you want to get mail you don't have any other choice. I've gotten it to work using WINE but then its even more unstable than usual. Sigh.
  • by gnarphlager ( 62988 ) on Monday January 10, 2000 @04:40AM (#1387401) Homepage
    There's a bit on the NY times website [] about this too.

    Personally, I'm not sure I really see the significance of this. Big blue likes linux. Okay, fine. They'll sell servers running linux. Neat. But what are they going to do to develop linux? Are they going to contribute to the community, or are they just going to make a quick buck on everyone else's work without having to worry about NT licencing fee. Don't get me wrong; it's great to see more linux servers and workstations in the world, and any exposure is good exposure, but if anyone has the means to help development, it's IBM.

    Too many corporations are looking at linux as a finished product, rather than a work in progress. It's not.

Disraeli was pretty close: actually, there are Lies, Damn lies, Statistics, Benchmarks, and Delivery dates.