Slashdot Mirror
IBM Calls Linux "Logical Successor" To AIX
pknoll writes "Though it probably won't happen soon, IBM is talking about Linux eventually replacing AIX. The article at Globe Technology states there are IBM folks working on 'chips for 2007' systems, and the viewpoint projected is described as 'multidecade,' but it's an interesting view into the future of IBM and Linux."
Are you kidding? What are those commercials about linux on TV then?
I think you are badly mistaken about the role that IBM wants to take with Linux. They aren't interested in putting Linux on the desktop for the same reasons they never put AIX on the desktop. For them (and for most people) its a server OS.
I agree it would be nice for them to push it on the desktop as well, but saying they don't support it just because that isn't the role they are taking with it is irresponsible.
The Anti-Blog
'cause AIX is actually pretty damn good. It may not have the rampant 'coolness' of Linux with all its little gadgets and what not, but it's a rock solid stable system with many advanced 'enterprise level' features.
In most of the ways that matter, AIX is well ahead of Linux. Seriously, Linux has some catching up to do if it to replace AIX.
What about the IBM Linux Technology Center.
What about the many, many people IBM donates to work on strategic open source initiatives.
What about nearly every IBM application running on Linux.
What about nearly every piece of IBM hardware running linux.
What about billions of dollars of services contracts to push the kernel's and distributions's limits, as well as keep places like Red Hat and SuSE alive with big fat checks from service contract customers buying their wares.
And what about the fact that just by saying 'linux is the future' IBM is making linux the future in the minds of a lot of people.
What did you eat today? http://www.atetoday.com/
10 years ago... people would ask
5 years ago
2-3 years ago
NOW :- We want to use linux, to reduce our IT budget cost . Also we are fed up with the security issues with M$ products and the licnesing costs are killing us. Plus we have heard that linux is an excellent replacement for legacy *inxs.
Although IBM may not have contributed directly to kernel code, they are doing a lot to improve LINUX's image in the mindset of MANAGERS of IT Project,
As they say, Win the MANAGER and the staff will follow.
for the last time people, I am "frodo from middle eaRTH", not "middle eaST".
Performance in the data center isn't there yet for Linux. Almost nobody with a serious databasee will run it on Linux. Even though DB2 and Oracle run on Linux, it's just not as fast yet. 2.6 should hopefully change this, as the kernel developers have been taking suggestions in this area to heart.
"Ignorance more frequently begets confidence than does knowledge"
- Charles Darwin
I'm going to try not to make this sound like a troll... but it's hard to be politically correct while looking a decade down the road....
Like IBM, SGI is also kinda-sorta planning on moving entirely to Linux in time. This makes me wonder what the long-term path is...
Once upon a time IBM and SGI were working with oldschool AT&T SysV Unix and BSD Unix, after years of tweaks, overhauls, and rewrites, each company ended up with their own distinct version of Unix. Obviously this won't happen immediately with Linux, but I would venture to guess that there will be significant forking over time. Right now SGI is using a slightly modified version of Red Hat 7.2 on their Altix machines (basicly Red Hat plus the patches from their "ProPack" overlay). As time goes on I would almost bet that the long term goals of IBM, SGI and others will not match up to those of RedHat and other distro builders. I have a feeling that, oh, maybe 10 years down the road each major big iron builder (IBM, HP, maybe SGI and Sun) will have their own distinct (and somewhat "weird") version of Linux.... and soon the term "Linux" will be as generic as "Unix".
This makes me wonder.... why bother with the Make-Work of moving to Linux in the first place? Why no keep working on the existing tuned kernels of AIX, IRIX, Tru64, etc?
IBM IS NOT IN THE LINUX DISTRO MARKET.
They are about embracing, and extending current technologies. This includes supporting Red Hat AND SUSE (heaven forbid they work with more than one linux distro)
And, for the UMPTEENTH time, IBM IS NOT OUT TO WIN LINUX ON THE DESKTOP. Last I checked, there aren't very many people out there running AIX on a Thinkpad, or a Desktop machine.
IBM is focused on the SERVER market with Linux, not the desktop market. There isn't a desktop market for linux at this time.
As a rock-in-roll Physicist once said, No matter where you go, there you are.
I work for a small company in Austin TX that ports software. We like to brag that we can port anything to anything, but in reality, all the work that I seem to be doing is porting from either Solaris or HP-UX to Linux. AIX takes a very close second to the targets that we are porting to. Of course, this is very biased, since we're an IBM business partner. :-) I'm sure there's ports going on somewhere to Solaris and HP-UX.
In all of the arrangements that I've been involved with IBM on, their people have been completely indifferent about porting to Linux in preference to AIX. They simply don't seem to care what the hardware is running, as long as the customer is buying shiny new IBM boxes.
Something interesting though - IBM's Visual Age for C++ compiler was a pain in the ass to figure out. There's a zillion command line switches, and getting the right ones set to build proper dynamic libraries took a bit of figuring out. gcc was much nicer in that regard. But, now that I've got them figured out, I really like IBM's compiler more than gcc 2.95. I haven't had the luck of using gcc 3.2.1 yet (third party libraries aren't typically built with it yet, and I use Debian at home) but I can't wait. That new gcc compiler will really be sweet.
If tits were wings it'd be flying around.
They aren't interested in putting Linux on the desktop for the same reasons they never put AIX on the desktop.
Crap. I better ditch my 400MHz Power3+ RS/6000 then. Since it's a server, I should certainly not be using it daily as my desktop. Damn shame to have the fast, pretty graphics card in it.
I've never thought that there was any mystery in IBM's interest in Linux. Their product line has no common OS. Linux turns the MVS, CMS, OS/400, AIX, Windows, etc., etc. etc., muddle of completely different operating systems for every flavor of hardware into something intelligible: We run linux top to bottom!!!
i le/368660.stm
It has added bonuses too:
It weakens Microsoft's operating systems monopoly
It gives IBM another crack at selling their apps on hardware MS would own if it ran Widows
It might even be payback for making IBM pay significantly higher royalties for Win95 than other large customers (as payback for OS/2 & Lotus Smartsuite). http://news.bbc.co.uk/1/hi/business/the_company_f
Gives IBM a chance to sell system integration services and service contracts.
Provides programmers world-wide to contribute to IBM's success
Gives IBM a story that sounds similar to Sun's:
Sun: Complete binary compatibility from Desktop to Midframe.
IBM: You can run linux top to bottom
No, I don't think that there is any surprise in this at all.
At this point, this is probably just a statement about likely future direction, and as such it doesn't mean much, but in the long run I would expect that many of the AIX engineering and support people can be retargeted for Linux. AIX has a lot of support for things their customers really need, and it will take a while to move the important bits of this over to Linux. Probably, they will not OS all of this, but it may become available for purchase for other platforms. All of this is good for the industry.
Not now. The Linux kernel, while improving, does not have the enterprise features needed by those who use AIX. That said, I do believe Linux can replace the AIX kernel. If and when it does, you still have to possibly port some of those tools. For example, AIX has some great commands (not just smit) that are very nice. Ones that come to mind are lsdev, lsfs, lsvg, lspv, the odm itself(not as bad as you think), lsattr and many others. Smit is a lifesaver when you just can't remember the commands to do a certain thing. Also, AIX's ability to expand filesystems on the fly, the LVM, HACMP, SP and other things are essential for AIX shops. The pSeries machines (otherwise known as RS/6000) are IBM's best selling servers. The Regatta (p690) is doing extremely well. Almost everyone I have come across who runs AIX (except us....we're cheap, er poor bastards!) has a p690. Also the Shark (Enterprise Storage System) is tremendous. In 2 racks you get redundant storage. One rack can die and the other takes over. Each side has it's own battery backup, plus there's 348 MB of Non Volitle storage. Also you can have 22 TB of SSA in that rack! All of that works because of the fine work IBM has done on AIX. Linux can replace it, but it will be a while! :)
Gorkman
First, let me point you towards Wikipedia's history of Unix
Also, I found this on Usenet, hopefully it will give you a good idea on the background of Unix:
And finally, correct me if I am wrong, but it is assumed that Linux gets it's name because it is the brainchild of Linus Torvalds
Actually you got it backwards, Multics was extremely multiuser, was the reasoning behind the name. UNIX started out as largely a singleuser simplification of Multics (which was a very complex system), first named UNICS, which makes the pun much simpler to understand. The name was changed to UNIX when multiuser support was added.
Multics is really to operating systems what Algol was to programming languages, the huge research project where they found a lot of the do's and dont's in operating system design. A great system all in all, a bit too ahead of its time though.
See http://www.wagoneers.com/UNIX/City-U/Multics/ for more information.
"Plus we have heard that linux is an excellent replacement for legacy *inxs."
I dunno, I think the first couple of INXS albums rock harder than any Linux distribution ever could.
So far, he's refused an absolutely huge patch from an IBM team, there were two competing implementations and the IBM team lost out (LVM?). Anyway, the team acted with dignity and honour, and got to work immediately with helping the other team. So - they have no history of this whatosever wrt to Linux.
Anyway, the moment there was a patent problem with the kernel, there'd be a non-infringing implementation in within seconds.
It is not that the Linux you know now is going to replace AIX. The point is that IBM will invest to bring Linux up to the level of AIX over the years, and when Linux is as good as AIX at what AIX is good at, you will get a product like AIX but with the "coolness" of Linux.
IBM is merely reinforcing their already rock solid commitment.
Sdelat' Ameriku velikoy Snova!
"And they get to milk developers who work for free!"
"Milk" implies that they don't contribute their share and from watching linux-kernel I can tell you thier engineers are responsable for many of the scaleabillity improvments added so far during the 2.5.x development cycle.
Check out the OSDL Data Center Linux project, as well as the Carrier Grade Linux projects.
Some good info about specific kernel changes can be found in the DCL Road Map.
OSDL has been *great* with regard to bringing companies and kernel hackers together on this subject. Gathering specs, performing QA, and allowing use of hardware. I had talked with Timothy Witham at Linux World about database performance, and he said he's seeing dramatic speed improvements on a 2.5 kernel over 2.4 (20%->30%, using standard TPC benchmarks). Though I don't remember the database he was using (not MySQL or Postgres, but it was open source I believe).
"Ignorance more frequently begets confidence than does knowledge"
- Charles Darwin
"Why no keep working on the existing tuned kernels of AIX, IRIX, Tru64, etc?"
Well, why don't IBM and SGI just stick with their old 'nixes then? If you fork Linux, you go into a full loop and have the same type of problems you had with old 'nixes. The beauty of Linux is not what it has become, but where it will always be going. Fork it, and you lose the most important "feature." I certainly think that each company will be using "niched" versions eventually (think, Debian vs. RH), but I really can't see why they would find it economical to fork from the main branch and return to the dark ages.
Linux provides an open standard that allows anyone (individual or corporation) to contribute to a standard without the fear that they will not be able to benefit from those contributions or, worse, have those contributions be used against them.
Maintaining an OS is becoming way too expensive, unless you have a near perfect monopoly and can control market prices. Smart companies will start to treat the OS as a "standard" (like HTML) and forget trying to control it. Instead, they will focus on building powerful applications to run on this "standard" OS, without fear that the owner will eventually decide to compete with those applications, leveraging their control of the standard, since there is no one owner in existence to control the OS.
It is my opinion that IBM is one of the smartest companies is existence these days. . .
Sdelat' Ameriku velikoy Snova!
The article points out that AIX is handled by the Server group at IBM, not the software group. So while this Mills guy says exciting things, he isn't necessarily the guy to make that decision.
Excellent point. Anybody who actually read the article (and it's been up on News.com's website for a couple of days now) knows that IBM's AIX folks are surprised to hear that AIX's days are apparently numbered.
Basically, the article quotes one guy from IBM as saying that he foresees the day when Linux will replace AIX in IBM's lineup. The odds are that he's right simply because it costs so much to develop a Unix and keep it current, and IBM wants to be able to have you scale up from a low-end Intel box to a Z-series mainframe with any stop in between and take your software with you. Linux is the one OS that runs on all of IBM's hardware.
But that said, it'll be awhile and the AIX guys won't go quietly. They'll probably have some kind of AIX-compatibility libraries that they'll license to their customers the way SCO is planning to do with their libraries. IBM may also port their AIX management tools to Linux and license those separately, as well. Who knows what the future will hold, but it's likely that Linux will simply absorb AIX's capabilities in IBM's product lineup at some point. This means that even if AIX goes away, it won't really go away--it'll just change shape.
One last point. As someone pointed out in the article, "IBM has never decommissioned an operating system, and they're not about to start now."
As we move forward, there will be variants of the common OS code base for different platforms and applications and, certainly, more applications and GUI's than you can shake a stick at, but the OS qua OS is pretty much finished. This is actually a good thing. It will lead to a stable platform for development of applications while freeing up OS kernel folks to actually do something new and different.
That is all.
So what if AIX has advanced features that Linux lacks? AIX market share is nothing compared to Solaris and HPUX, and most CAD/EDA Unix shops running Solaris/HPUX have found that Linux on Intel handily outperforms Solaris/Sparc & HPUX/PA-RISC on pretty much anything that doesn't need 64-bits or more that 3GB RAM. So if Linux already does well on the low end and will almost certainly do well in the high end, why shouldn't IBM get in early and position themselves as a leading Linux vendor? It's not like AIX is going to suddenly turn around and outpace Linux usage.
I applaud IBM for realizing AIX had it's chance and won't be a dominant player inteh OS field. They'll roll that AIX expertise and technology into Linux and the whole Unix world, IBM included, will be better off for it.