|WikiProject Software / Computing||(Rated Start-class, Mid-importance)|
- 1 Unix transition guide
- 2 AIX 1.0 and SVR3
- 3 AIX on IBM Mainframes
- 4 Probably a Silly-Mistake
- 5 A Correction
- 6 "but also beat the competition by a factor of 10 in floating-point performance."
- 7 Incorrect info about SMIT command log
- 8 Article name "IBM AIX (operating system)"
- 9 SCO lawsuit
- 10 Journaling
- 11 Pronunciation?
- 12 Fair use rationale for Image:Aix logo.gif
- 13 AIX on Bull Express5800
- 14 Article name
- 15 PowerPC G4?
- 16 Multics family
- 17 Screenshot
- 18 The Coherent file system
- 19 Requested move
- 20 Is AIX still being developed?
- 21 SCO vs IBM in the lead
- 22 Korn shell
- 23 IBM AIX PS/2
- 24 Picture
Unix transition guide
Anyone care to outline a introduction guide for persons already familiar with Unix? Links? Thanks, Gchriss 20:37, 5 August 2005 (UTC)
AIX 1.0 and SVR3
AIX on IBM Mainframes
I changed the previous MVS/ESA section and replaced it with AIX on IBM Mainframes. AIX/370 and AIX/ESA information is very hard to come by because the operating system was not much of a success and didn't last long. I also removed the note about AIX/ESA being based off of OSF/1 because I could find no credible information saying such (and it seems unlikely). I found the introduction date on IBM's Year in Review website and based the release date off of Unix History (http://www.levenez.com/unix/). I also found some information on the MVS/ESA article. There are official IBM documents in BookManager format on AIX/370 available at this site (http://www.tavi.co.uk/ps2pages/aix.html). TimP 21:27, 14 January 2006 (UTC)
As someone who worked on AIX/ESA, I can confirm it was indeed a port of OSF/1. Why do you think it's unlikely? Also, the information in the AIX article that AIX/370 was rebranded as AIX/ESA is wrong. They were separate operating systems. Finally, while AIX/370 was restricted to running as a VM guest, AIX/ESA did not have this restriction and could be run natively. -Dave Marquardt, AIX/ESA developer from 1989 to 1993.
AIX/ESA did run natively on the mainframe and the kernel had little in common with AIX for the RS/6000. It was a port of OSF/1. I also was a developer of this OS from 1989 to about 1992. My team did the disk and tape support (which was ALL native). I have seen an AIX/ESA machine with a 1000 people logged on. Technically impressive, but didn't sell! Some said this was because MVS Open Edition was developed as a corporate "immune response" from the very large MVS development team. - Dick Johnson
AIX/370 was a port of the LOCUS clustering operating system from Locus Computing Corporation, which was a commercialization of an ARPA research project at UCLA led by Dr. Gerald J. Popek (of Popek and Goldberg virtualization requirements fame). You can see the book on Amazon.com -- ISBN 0262161028, "The LOCUS Distributed System Architecture", edited by Gerald J. Popek and Bruce J. Walker, MIT Press, 1985 (you may notice my name on Chapter 4). It was based on 4.1BSD and was not anywhere near the same branch of the UNIX family tree as IBM's AIX on the RT, for example. The same OS also ran on the IBM PS/2, as well as the VAX and a couple other Instruction Set Architectures (same source, different binaries of course). I think IBM called it Transparent Computing Facility (TCF), and I think it was also called Transparent Network Computing (TNC) at some point in time, I believe by Locus attempting to sell the technology to other buyers in addition to IBM. (It is a bit disappointing to note that the IBM370/AIX page does not seem to recognize this, and talks about the RS6000 version of AIX and the IBM370 verion as though they were the same thing; but then it seemed to me that IBM was pretty deliberate in using the "AIX" brand in a way that was consistent and made it look like one product.) The book referenced above was published during a time when the contract between Locus and IBM was still supposed to be secret, but Locus apparently got permission to credit IBM without making a direct statement, and that can be seen on pages xv (Preface) and xvii (Acknowledgments). I think it was an editing error that the example discussed on page 104 (first paragraph of section 6.5) was published talking about a "370" site -- I think all those examples were supposed to have been changed to "68K" instead (which LOCUS did run on). Amazon.com will let you see these -- I used the "search inside" feature to find "IBM" and "370". There are a few hints around: http://www-sop.inria.fr/parallel/DR:/lsf/man/tcsh.1.html (search for "TCF"); http://www.puffin.com/~lynn/99.html, http://www.garlic.com/~lynn/2001f.html#22, http://www.garlic.com/~lynn/2006c.html (search for "locus") Dave Butterfield 23:25, 20 September 2006 (UTC) co-chief technical lead for support of AIX/370 and AIX/PS2 at Locus, February 1989 - May 1992.
I just looked around and found that Locus delivered AIX for both the 370 and the PS/2 to IBM for acceptance testing in October and November of 1986. It was delivered to IBM in Boeblingen, West Germany by Evelyn Walton, one of the founders of Locus. Dave Butterfield 00:56, 21 September 2006 (UTC)
I'm no expert on this, but I think this section should mention the current Unix subsystem on z/OS. If I'm right many of the current newer IBM mf products run in this userland. -- Eric Apse —Preceding unsigned comment added by 220.127.116.11 (talk) 23:08, 3 June 2009 (UTC)
- Only if z/OS UNIX Services are directly related to AIX, and I'm not sure they are. Letdorf (talk) 13:33, 4 June 2009 (UTC).
Probably a Silly-Mistake
Last line of the Text says:
"If you are on a text based terminal, running the smit program will invoke the text-based version."
Shouldn't it be:
"If you are on a text based terminal, running the smit program will invoke the GUI-based version."
Ignore the post if I am wrong, as I have no AIX experience. Just a curious reader and Linux/Unix enthusiast, who thought it is probably a mistake. Thanks. --18.104.22.168 09:33, 24 January 2006 (UTC)
I just tested and running "smit" on a text based terminal does indeed bring up the text based version. I to make the hard call to the text based version the command is "smitty". 22.214.171.124 21:09, 10 March 2006 (UTC)
I am pretty sure the smit command senses if you are running in GUI or text mode and will use the appropriate display.--Mrmouse 15:40, 4 July 2006 (UTC)
This is correct. If you invoke "smit" without any options, it will determine if the $DISPLAY variable is set; if so, it will invoke the GUI-based version; if not, it will invoke the CLI-based version. Invoking SMIT as "smitty" forces the "tty" (CLI-based) version to run. This is, of course, by design. --SolarisBigot 16:55, 23 July 2006 (UTC)
Actually AIX stands for "Advanced Interactive eXcutive" if memory serves me well. I remember back in the version 1 days they used to print that phrase on the 8" floppies you used to build it, and that always stuck wtih me. 126.96.36.199 21:05, 10 March 2006 (UTC)
I'm not sure what is meant by "AIX v3 was the first OS to introduce a journaling file system". Does this mean it was the first to natively support a journalled FS? Because Veritas_File_System was the first journaling FS.
- AIX was the first to naitvely support a journalled FS. AIXv3 shipped in 1990 (and was working in-house years prior to that). When did Veritas start shipping thier product? linas 00:38, 6 July 2006 (UTC)
"but also beat the competition by a factor of 10 in floating-point performance."
does anyone else find the breathless enthusiasm for AIX and the RS/6000 a little POV? i mean, just a bit?
Flagged appropiately. --SolarisBigot 17:09, 23 July 2006 (UTC)
Also in that mindset, the claim "not only outperformed all other machines in integer compute performance". I suspect that "all other machines" here, like "the competition" above, are categories of convenience -- they include only what machines they must in order to make the claims true, just as integer and floating-point performance are defined by whatever benchmarks make the claims true. As I recall, the RS/6000s really were speedy machines -- I had one on my desk -- but this is used-car-salesman hype at its deepest. 188.8.131.52 (talk) 19:26, 11 July 2012 (UTC)
- I don't think it was POV at all; you can look up the numbers on the chipsets of the era, and compare them. Dinner table discussions were about competitors being driven out of business. The competitors here were SGI, HP and Sun, to a lesser degree Apollo, Intergraph and a few others I can't remember (E&S Evans & Sutherland ??). (although I guess Apollo had already gone down by then). They were competitors because the were selling Unix workstations at similar price-points: in the $10K to $100K range, with roughly similar capabilities for graphics, RAM, disk storage, etc. What IBM had done with the floating point unit had never been done before: the snobbery of the time was that "Unix computing was all about integer performance", that this was the only thing that counted. You could see this in magazines like Unix Review (?) (or was it Unix Today (??)) that more-or-less pooh-pooh'ed the importance of floating point. This was bizarre, to say the least, it seemed to completely miss the point that much/most of the market was scientific computing, and that actual customers were running numeric codes on these machines! So when IBM came out with a floating point pipeline (an FPU) that was wired into the integer pipeline, instead of being a co-processor, that was a "game changer", as they used to say. The fact that the integer unit was equal or faster than anyone else's helped deliver the launch. Of course, it turned out the workstation game was a game of "leapfrog": it wasn't long (typically 6-12 months) before someone else announced a "the fastest" machine, with some gamesmanship of comparing apples to oranges. Not at all unlike the smart-phone market today. Please note: cell-phone chipsets are designed by many of the same engineers who worked on these machines back in the 1980's & 1990's: you can walk the Qualcomm hallways and bump into the same engineers who worked at IBM, SGI, etc. back in the day... in some ways, cell-phones really aren't all that different, just 1000 times smaller, cheaper, and faster. linas (talk) 20:40, 26 August 2012 (UTC)
- The problem seems to be that, even if the RS/6000 machines took the lead among their peers from other vendors, that's not the claim that's being disputed — the article claims that the POWER machines “outperformed all other machines in integer compute performance”, where “all” would include multiprocessors and supercomputers. If the proper comparison is between machines with equal numbers of processors, comparable quantities of main memory, and similar list prices, then perhaps the overly-enthusiastic text would benefit from some of those qualifiers. :) 184.108.40.206 (talk) 23:48, 23 November 2012 (UTC)
The HP snake, and later the DEC Alpha, did beat the RS6k on arithmetic performance provided all data was in the cache, some 256kb. The RS6k had at least twice the memory bandwidth of its early competitors. It did very well on large memory jobs, such as vectorized fluid dynamics code. The difference was the IBM direct mapped cache vs associative caches of its competitors. For example, circa 1992: IBM 550 did 18Mflops on triads, the Snake 712 did 6Mflops, and the Cray YMP was 289Mflops. Snakes and Alphas made nice personal workstations, while RS6ks were more likely to be found running a large batch job taking days of CPU time. Memory bandwidth stats can be found on the Stream Benchmark page at U. Virginia. It was the RS6k memory performance which prompted this project.220.127.116.11 (talk) 02:41, 2 November 2015 (UTC)
Incorrect info about SMIT command log
smit.log contains the output and results from actions that are performed, but does not contain much that would be usable in writing scripts. smit.script contains all the commands that are generated to execute an action, and is where one would get the commands to write their own script. -- Kel Byers, AIX Test Engineer
- I updated the article to reflect both files as you mentioned. --Unixguy 15:40, 24 October 2006 (UTC)
Article name "IBM AIX (operating system)"
Why there is "(operating system)" in the article name? I think that there is not anything else called "IBM AIX". I am going to suggest renaming the article. --pabouk 07:30, 30 March 2007 (UTC)
- I agree that this is redundant, but I don't think it's the (operating system) part that should go, it's the "IBM" part. Coherent with: Tru64 UNIX, Solaris Operating System, Mac OS X, HP-UX, OS/2 and IRIX. All proprietary operating systems without the vendor name in the article. The odd man out is Microsoft Windows but that's all in order since MS couldn't register just "windows" as a trademark. -- Henriok 15:21, 30 March 2007 (UTC)
- I agree that the "IBM" part is redundant. I was not sure whether it has not been part of the name. I also asked User:Rwwww who renamed the article for the reasons for renaming but I am not sure whether he will answer because he has not contributed to Wikipedia since November. --pabouk 15:29, 2 April 2007 (UTC)
- There ia a long response on my talk page. tooold 14:28, 10 June 2007 (UTC) (Rwwww)
- I think that as long as AIX operating system redirects to the main page IBM AIX (operating system) then we are ok. A person would still arrive at the end page without knowing that IBM is part of the name. Here are all the current redirect pages:
- --Unixguy 17:04, 11 June 2007 (UTC)
SCO has not claimed (in court) that IBM misappropriated code from sysV and put it in AIX (or anywhere else), but rather IBM took code that IBM wrote for other operating systems, and ported that code into AIX, and then later on ported that same code into linux. What SCO says to the media is sometimes different from what they say in court. 18.104.22.168 18:40, 2 April 2007 (UTC)
This assessment it is not true: AIXv3 innovated in several ways on the software side. It was the first operating system to introduce the idea of a journalling file system, JFS, which allowed for fast boot times by avoiding the need to fsck the disks on every reboot —The preceding unsigned comment was added by 22.214.171.124 (talk) 09:36, 24 April 2007 (UTC).
- I have absolutely no idea what's true. We cannot replace something that might not be true with something else that might not be true either. At least I need some evidence before I change anything. Henriok 09:55, 24 April 2007 (UTC)
- I added the citation for the JFS claim. The source is primary (IBM) but at least the assertion is documented. ChrisLS120 (talk) 05:33, 10 February 2011 (UTC)
I stated at v3. The partitioning system was LVM. Wasn't AIX the first to use LVs exclusively? Volumes could be expanded on the fly, but not shrunk. It also used fixed-size loop or cyclic files to store auditing data.126.96.36.199 (talk) 02:47, 2 November 2015 (UTC)
How is AIX pronounced? I think the article should tell us: is it a single syllable "AYKS", or is it two syllables like "AY-iks"? The latter I thought was the case, to identify it as part of what we now call the *nix family of operating system - albeit without the "n" in this case. Are there any sources that say one way or the other? Is it still lodged in the memories of those who worked with it? --Nigelj 22:13, 25 August 2007 (UTC)
- "Aye-Eye-Ecks" is the way I've heard from representatives from IBM. Listen to IBM's developerWorks podcast to hear for yourself. -- Henriok 14:50, 27 August 2007 (UTC)
Fair use rationale for Image:Aix logo.gif
Image:Aix logo.gif is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.
Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Wikipedia:Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.
If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images uploaded after 4 May, 2006, and lacking such an explanation will be deleted one week after they have been uploaded, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.
BetacommandBot 22:30, 29 October 2007 (UTC)
AIX on Bull Express5800
I'm not entirely convinced by the cited website mentioning an IA-32 port of AIX 5L for the Bull (NEC) Express5800. Does anyone have any other references for the existence of this? Letdorf 13:19, 8 November 2007 (UTC).
- Bull sells rebranded IBM pSeries with AIX and is also involved in the development of AIX (on POWER) but that's the only link I'm aware of, so the claim of an IA32-port of AIX done by Bull is IMHO highly doubtable. --Kvedulv 14:12, 8 November 2007 (UTC)
- Removed per talk. /Blaxthos 18:29, 8 November 2007 (UTC)
Anyone please explain this to me: why the name of the article says "IBM AIX"? Looking at similar OS articles, such as IRIX and HP-UX, the developer company isn't mentioned in the title, so should not it be here. Or, otherwise, those articles aforementioned should be renamed to "SGI IRIX" and "HP HP-UX" (yes, double "HP" here, since "HP-UX" is itself a name/trademark) to meet the common sense consistency requirements. — 188.8.131.52 (talk) 18:11, 15 October 2008 (UTC)
- IBM call it "IBM AIX" as per this page and many other sources. Not in prose but in headlines. As should we.-- Henriok (talk) 08:56, 13 October 2008 (UTC)
ج ح خ د ذ ر ز س ش ص ض ط ظ ع غ ف ق ك ل م ن ه و ي —Preceding unsigned comment added by 184.108.40.206 (talk) 06:47, 8 January 2009 (UTC)
- Please keep the discussion in the one place. It started in Talk:Mac OS X. AlistairMcMillan (talk) 18:40, 16 September 2009 (UTC)
I added a screenshot of an AIX 5.3 login, but it was speedy deleted as a violation of IBM's copyright.
- Added it back now
The Coherent file system
Uh, i can't seem to find a suitable spot to mention the early Coherent filesystem ; perhaps someone else could ? A few sources :
- file:///usr/src/linux/Documentation/filesystems/sysv-fs.txt (if you're on a Linux system)
- Unfortunately hardly anything in http://tldp.org/HOWTO/Filesystems-HOWTO-9.html#ss9.26 , except that it's classified as a System V FS derivative (in agreement with the previous ref)
- What's the connection between Coherent (operating system) and AIX? Letdorf (talk) 13:09, 24 March 2010 (UTC).
- Aha, thanks, i didn't know of that OS. To answer your question, here is, from the 1st source i mentioned :
These filesystems are rather similar. Here is a comparison with Minix FS: * Linux fdisk reports on partitions - Minix FS 0x81 Linux/Minix - Xenix FS ?? - SystemV FS ?? - Coherent FS 0x08 AIX bootable
- This is very little indeed. So perhaps i should rephrase my original question : which file system was originally on AIX, prior to JFS1 ? and come to think of it, is JFS (2?) the only file system you get on AIX, besides the networking FSes ? So that means that everything is journalled ? In which case, isn't that overkill ? Etc. --Jerome Potts (talk) 02:05, 26 March 2010 (UTC)
- That's a slightly confusing list. What I think it is saying is that, on PCs, AIX (presumably AIX PS/2) and Coherent used the same partition type code in the BIOS MBR for their filesystem partitions (although this web page says the code was 0x09, not 0x08). That is probably coincidence and I doubt the two systems have anything else in common. As for the native AIX filesystem before JFS1, I'm not entirely sure what AIX/RT, AIX PS/2 or AIX/370 used, but it seems AIX/6000 always had JFS - AFAIK the first (customer) release of AIX/6000 was v3.1. Letdorf (talk) 13:29, 26 March 2010 (UTC).
Is AIX still being developed?
That latest stable release is almost 3 years old, which is a long time in computer language terms. Is there expected to be another major release? —Preceding unsigned comment added by 220.127.116.11 (talk) 16:15, 1 September 2010 (UTC)
- Well, AIX is a stable platform. They are betatesting 7.1 as we speak. Read about it here: http://www-03.ibm.com/systems/power/software/aix/v71/index.html -- Henriok (talk) 22:30, 1 September 2010 (UTC)
SCO vs IBM in the lead
Does the SCO vs IBM really belong in the lead? I just read the article for the first time and, while interesting, I can't see that SCO vs IBM is really an important part of an article on AIX that would belong in the lead. Also, much of the material about the court case only appears in the lead, whereas the lead should summarise the article. Shall I move it? Alex Harvey (talk) 05:02, 2 May 2011 (UTC)
- Yes, not really lead-worthy. Letdorf (talk) 13:22, 2 May 2011 (UTC).
AIX uses the Korn shell as default, which is probably special enough to mention here.  Does anyone know since when this is the case? The Korn shell wasn't around in 1986. FuFoFuEd (talk) 12:06, 30 May 2011 (UTC)
IBM AIX PS/2
I worked in the Danbury,CT, lab where AIX PS/2 was developed, and can offer some insight and possibly corrections to other sections here, and perhaps the Wiki itself. My initial role was as a member of architecture team where I was a *nix newbie but owned the Final Programming Functional Specification (FPFS). I was responsible for integrating the architecture team members' content, distributing the FPFS across the business and securing concurrence. I got that job because my previous project (VideoText/370) had been cancelled, I lived nearby (no relocation expense) and I knew ISIL (BookMaster).
In a nutshell, AIX PS/2 was to integrate SVID(TM) and BSD 4.3(TM), and add things like DBCS, X-Windows, a "DOS Compatibility box" (from Locus), etc. It was intended from the beginning to become the source base for implementations of AIX planned for the RS/6000 and System 370, so those organizations were stakeholders from the FPFS phase through General Availability (GA). Our GA was pegged to the PS/2 GA to ensure there was an "IBM" (i.e., non-Microsoft) OS available for that platform from day 1. Interestingly, our device drivers were based on work from a Boca lab that had taken over driver development for OS/2 when MS said "seeya!".
Subsequently I lead the comnmand port team (mostly contractors), mainly in normalizing interfaces, big- and little-endian issues and DBCS support. After that I led the LPP test team, and once we hit SVT I prepared materials for and ran the Release Manager's daily "must-fix" meeting, where stakeholders from Austin (RS/6000), Kingston (370), Palo Alto (X-windows), Santa Monica (Locus) and others regularly participated.
So much for bona fides.
My issue is with the characterization of Locus as the "developer" of AIX PS/2, and possibly, AIX/370. I can only speculate on the latter, but as for AIX PS/2 they contributed only the "DOS Compatibility box" which allowed DOS applications to run on a PS/2 under AIX. This was a technology they marketed separately for other *nix variants, and it worked pretty well - I spent a month there during functional test (all the servers were named after LOTR dragons). If I remember correctly, they released a new version shortly after AIX PS/2's release that contained a number of ennhancements absent from the version IBM had licensed. If anyone that was at Locus at the time can comment I'd be interested to hear what they have to say, but these are my recollections.
As for AIX/370, I know that Mike Schmidt (lead AIX PS/2 architect) went to work on that project in Kingston, and AFAIK the code base was our frozen code from AIX PS/2 R1. It is entirely possible that Locus contributed clustering support, but I have no direct knowledge of that.
We made GA, by the way, though the efforts of a lot of great people, with whom I've unfortunately lost contact - people like Mike, Tery McAuliffe, John Spannaus and Bill Jones, who I follwed to a new IBM gig in Tampa (and later another gig with a different comany; he has since retired). And especially my LPP team go-to, Lisa Wright, whose mottos were "Do it NOW" and "That's not good enough" (exclamation points would be redundant). May every team have a Lisa. — Preceding unsigned comment added by 18.104.22.168 (talk) 08:59, 5 February 2012 (UTC)
- Hi, thanks for your comments. I think the mention of Locus in the article, in relation to AIX PS/2 and AIX/370, was based on the comments above by Dave Butterfield. If you can find some reliable sources that shed more light on the origins of AIX PS/2 or AIX/370 and could be cited in the article, that would be great. Regards, Letdorf (talk) 19:41, 6 February 2012 (UTC).
Speaking as the kernel development manager for AIX PS/2, I can assure you that Locus did the lions' share of the initial kernel development work. My recollection was that AIX started out in Danburry with the "IBM Instruments Computer", but when they wanted a PS/2 port they asked Locus to do it. It was Chimerical port of Locus' 286 Transparent Distributed Computing Unix, and the AIX High Function Terminal, to which Merge (led by Dave Butterfield) was added. The port was, in my mind star-crossed. IBM (with Mike Schmidt as lead architect) wanted a team with serious Unix kernel chops to do a straight SV 386 port, but with an homage to the IBM Instruments Computer port. Locus wanted a vehical for getting their heterogenous distributed operating system into the market. These three conflicting goals led to much architectural and managerial strife, and I don't believe anyone really got what they wanted. My role in AIX/370 was only peripheral (I was the architect and kernel lead for VM/IX, had long struggled with AIX/370's predecessor, and strongly argued the necessity of the new port) so I will not comment on that port. If you have more specific questions, I still know most of the Locus/Interactive principals. — Preceding unsigned comment added by MarkKampe (talk • contribs) 14:52, 17 February 2014 (UTC)
@Henriok: what, in your opinion, is the added value of the image
given that we already have a picture of a login prompt for an older AIX, above? (Note that the other one uses green text, which makes it a bit more of an eyecatcher; I selected which one to remove on purely aesthetic grounds.) QVVERTYVS (hm?) 17:22, 26 February 2016 (UTC)
- There are actually four pictures showing the CLI in AIX, which is prudent when describing a UNIX operating system largely driven bli CLIs. Minute differences between versions are interesting and I'd actually would like to have more, showing more versions and platforms. Removing pictures that actually convey information based on aesthetics are not a valid reasen in my opinion. If you want to remove pictures, you should probably remove some (all?) of the three pictures showing hardware. This is an article about software, so removing images describing the software while keeping images that show something else is.. not the priorities I'd make. If it'd be up to me though, I'd keep all pictures and add more. Henriok (talk) 11:28, 27 February 2016 (UTC)