back to article IBM shifts remaining US-based AIX dev jobs to India – source

The Register has learned that IBM has shifted the roles of US IBM Systems employees developing AIX over to the Indian office. Prior to this transition, said to taken place in the third quarter of 2022, AIX development was split more or less evenly between the US and India, an IBM source told The Register. With the arrival of …

  1. Snowy Silver badge


    Indian Business Machines?

    1. seven of five

      Re: IBM




      1. Peter Gathercole Silver badge

        Re: IBM

        Many years ago, when working in IBM in the UK AIX support centre, I had occasion to call AIX the American Interactive eXecutive (this was after several exchanges with a US based level 3 AIX support person who was saying that the A4 papersize was a mere niche size that nobody really used). I actually wrote this in the Problem Management Report (PMR) on RETAIN, and you weren't supposed to put personal opinion in PMRs.

        I guess I can no longer call it that!

        In the case of the problem described, the call was escalated up the management chain in the US to try to get me reprimanded, then down the management chain in the UK, and I was hauled in front of the IBM European Product Manager for midrange systems.

        After explaining the problem to him (I had fully diagnosed the problem, knew exactly what was happening, and devised a local fix, all of which was in the PMR and associated APAR), he said that my language was actually rather restrained for extent of the problem (which meant that you had to power cycle an IBM 4019 laser printer if you tried to print on A4 paper, even if you set everything up in the printer and the print for A4 paper. The fixed printer pre-amble the AIX printer interface sent did a sledgehammer reset of the printer back to factory defaults, and it then jammed because A4 paper was too long when it thought it was printing to US Letter).

        It went back up the UK management chain, and then back down the US management chain, and the problem was eventually fixed in three different places (printer firmware, the default AIX printer pre-amble settings and the piobe printer backend to allow it to be configured). So I guess I was justified.

        Fortunately, I don't deal with L3 and developers for AIX any more (at least in my current role), so I won't have to face the problems of describing a difficult problem to someone for whom English is not their first language,

        1. JacobZ

          Re: IBM

          "Fortunately, I don't deal with L3 and developers for AIX any more (at least in my current role), so I won't have to face the problems of describing a difficult problem to someone for whom English is not their first language"

          On the plus side, at least you would be describing it to somebody with a broader worldview than an American.

          You might also get exposed to some lovely contributions to the English language originating in that part of a world, for example "prepone", said when bringing a meeting forward.

          1. Peter Gathercole Silver badge

            Re: IBM

            I see what you said there. Touche!

        2. Anonymous Coward
          Anonymous Coward

          Re: IBM

          > I had occasion to call AIX the American Interactive eXecutive ... I guess I can no longer call it that!

          Americas Indian eXodus?

  2. man_iii

    Redundancy relocation

    When there exists SLES ppc64le and RHEL ppc64le .... Why Why would anyone want to continue with AIX ???

    Is it such a big impact if 80 AIX Dev positions got offshored?

    If you had said Redhat Devs' positions were moving to India that would ring alarmbells.

    Pirate becos it is offshore where bribery .... ahem..... I mean corruption .... errrr .... piracy.... arrr..... abounds.

    1. seven of five

      Re: Redundancy relocation

      > When there exists SLES ppc64le and RHEL ppc64le .... Why Why would anyone want to continue with AIX ???

      Your ignorance is my job security. Seen all three of them. Did not fancy what I had to endure, stayed with AIX. So do many companies.

      1. katrinab Silver badge

        Re: Redundancy relocation

        Especially given that IBM owns Red Hat ...

        1. seven of five

          Re: Redundancy relocation

          yes, now. I gave up on ppcle three years ago, so things might be changing (slowly).

      2. Anonymous Coward
        Anonymous Coward

        Re: Redundancy relocation

        Does AIX ("aches", around these part) have systemD?

        If not, then that's one thing it has going for it.

        But just the one.

        1. rnturn

          Re: Redundancy relocation

          > If not, then that's one thing it has going for it. ... But just the one.

          Heh. I still recall some older UNIX books having chapters or appendices devoted to the way certain aspects of the OS worked under AIX. Haven't touched the OS since the last employer that was running an RS/6000 and don't miss either at all.

        2. Peter Gathercole Silver badge

          Re: Redundancy relocation

          Systemd infects only Linux.

          No UNIX system has systemd.

          This is one of the reasons it's becoming more difficult to back port tools from Linux to traditional UNIX systems, along with dbus, udev and the /sys pseudo filesystem, as modern tools expect these things to be present.

    2. Peter Gathercole Silver badge

      Re: Redundancy relocation

      Although Linux performance on Power is actually better in some cases (a lot of the AIX toolset is still actually built as 32bit binaries, at least on AIX 7.1 when I last looked), and there have been performance bottlenecks in both the Network stack, and disk stack for the very highest loads (I'm talking HPC workloads, and this is from a few years ago), AIX remains quite popular in some niches.

      But the RAS features (Reliability and Serviceability) for AIX, and their interaction with the Power hardware is excellent, meaning that AIX systems just run and run with very few problems, and with good recovery options if there are problems. This is the reason many organisations still want to have AIX on Power in their system mix, even though it would be cheaper to run Linux on x86_64 once the migration has been done.

      If the Linux people want to really displace AIX on Power, this is what they need to aim at, but unfortunately, it would be quite difficult to get all of the Power RAS features mainlined in the Linux kernel.

      If IBM tries to maintain a private Linux kernel for Power systems to implement these features, then they start losing some of the commiditization that Linux brings, and actually start building the costs back into the support of the platform.

      The core part of AIX is functionally stable, and has been for more than a decade, but IBM keeps rolling out new capabilities, such as dynamic kernel patching, live partition mobility and fast system deployment in hyperconverged environments (yes, you can run AIX on Power in several Clouds), so I guess that AIX wil be with us for a while longer.

      I'm sure that Power as a platform will outlast AIX, but I think what may happen is that OpenPower systems, missing many of the RAS features to reduce the cost to better compete with Intel and AMD, will displace the current Power systems, to the detriment of the platform as a whole.

      1. Anonymous Coward
        Anonymous Coward

        Re: Redundancy relocation

        (Posting anonymously, as I'm still under Big Blue's wings). It's sad seeing how IBM has been undermining AIX year after year since I first set myself to learn its ropes, about 15 years ago. At one time I was marveled of how the Boulder development staff was keen to learn about new errors or performance issues and jumped in to actively engage in debugging and re-writing commands or code.

        I used to work for a Fortune 100 customer who had a carte blanche for AIX Enterprise licenses and a more than respectable testing environment, and we were encouraged to try new ways of configuring it (be it HA clusters, the then-new VIOS, or high speed network fiber cards). There was this L4 support guy I met every now and then when opening a PMR, we ended up being very close and we started chatting about our personal life (you need to kill those long minutes waiting for IPL when trying a new solution). Then I moved to a different account and maybe 4-5 years after I found he got assigned to my last PMR. I opened the chat by asking about his family (his younger son was bound for Harvard last time we spoke), and the person at the oher end of the Slack conversation knew nothing about what I was talking about. I pushed the envelope a bit more, and found out the guy I was chatting with was from Egypt, not the US, as my usual L4 support guy was. Yet they were using his name when answering problem calls.

        I guess the Powers That Be™ decided IBM i and mainframe were to become moar profitable, and the venerable AIX development team was first decimated and then relegated to the "B" league.

        1. Peter Gathercole Silver badge

          Re: Redundancy relocation

          I'm still providing service to IBM (as a contractor), but I use my name on posts that don't directly comment on IBM's actions. I do sometimes post anonymously if I feel the need to say something I rate as likely to be unpopular.

          Most AIX core development was done in Austin, not Boulder, although there were people working on AIX and associated software in many IBM locations, and several technologies have already been moved out of the US. Back in about 2014, I had a real difficult problem in the guts of the memory allocator in the large NUMA systems, and it was really difficult finding anybody in the US support teams who had the experience and knowledge to explain how it worked in order to overcome the problem.

          Not sure where you get L4 from. The support levels used to be L1 being someone who could talk your language and read the documentation, and would be your interface with the higher teams. L2 were people who had access to hardware and could attempt to reproduce problems, L3 was developer level people who were working on the problem queues to actually develop fixes, and above that were the development teams themselves. At the higher levels, people tended to move between L3 and development quite frequently.

          I was a technical team leader in an L2 support centre in the UK in the '90s just after the RS/6000 was launched, and in the 2010's I was in a very enjoyable gig that gave me pretty direct access to some of the development teams for AIX, GPFS, LoadLeveller, application development tools and TSM. This, fortunately, allowed me to mostly bypass the lower support levels, but I did have to actually get the PMRs raised, even if they were fast-tracked to L3 and above.

          Moving the primary support location for EMEA to Egypt happened mere months before the Arab Spring, and it actually became quite difficult to get a response from them during those troubles. Also, the people working the calls became very aggressive about closing calls, even before they had been fixed. Nowadays, I don't know who will answer a support call, not that I've had to raise any recently. I do know that some of the later support calls appeared to be handled from Hungary.

          I will be sad to see AIX further wither to a legacy OS, as it has been my primary technology for over 30 years, but as I'm merely a few years from getting my state pension, I probably won't have to watch it that much.

    3. JacobZ

      Re: Redundancy relocation

      There are some specific workloads where AIX performance is substantially better, e.g. traditional analytics.

      There are also some long-standing mission-critical workloads that absolutely nobody wants to touch, least of all to port from AIX's near-Unix implementation to Linux. (Somebody once explained AIX to me as "imagine you'd never seen Unix, but somebody had described it to you over a noisy phone line...").

      1. Peter Gathercole Silver badge

        Re: Redundancy relocation

        Hmm. That's a bit rich calling AIX a Near UNIX implementation! AIX is a UNIXtm, nearly the last one that owes anything to the original AT&T code (I'm excluding Solaris and HP/UX, as they're pretty much legacy now).

        I suspect that the comment you quote may well have come from a person familiar with Sun. In the mid '90s and 00's such people thought that SunOS/Solaris was UNIX, but in reality, SunOS was as much a flavour of UNIX as AIX was. You'd have to look at UNIXware to see something that was real genetic UNIX with little vendor change (yes, I am aware that Sun were part of the SVR4 development, but even there, Solaris implemented some things differently).

        It is true that IBM introduced differences, The actual UNIX system call layer described in section (2) of the manual was enhanced, often with extra parameters, but the version exposed to users under the default call names worked the same (often the system call name had an "x" appended to it to get the enhanced function, so open() gave the normal function, and openx() gave the enhanced function). But AIX remained SVID, Posix and UNIX03 certified, and is now compliant with the Version 4 of the Single Unix Standard maintained by The Open Group.

        In addition, many of the management interfaces were different, but they were not in the UNIX standards. But other vendors had their own management tools, IBM was not unique there.

  3. Reginald O.


    What does IBM do anymore? Anyone know? Seems like a horseshoe company that failed to become a tire maker.

    1. seven of five

      Re: IBM

      You'd be amazed how much money there is in horseshoes (and hoovenails).

      citation: Friend is a blacksmith. She lives rather good from it. And she is very busy.

      1. MyffyW Silver badge

        Re: IBM

        A horseshoe maker is rather a specific operation, being a blacksmith opens up many sources of revenue

        1. Dante Alighieri

          Horse shoes

          Farrier better to be an expert in a niche.

      2. disgruntled yank

        Re: IBM

        Not really what I think of when I hear "big iron", but good for her!

    2. Anonymous Coward
      Anonymous Coward

      Re: IBM

      "What does IBM do anymore? Anyone know?"

      Systems that run 24X7, with updates that are a walk in the park (ie. doesn't break every 10% of the executables running on top), good and powerful CPUs, well integrated into the AIX OS. Other stuff as well.

  4. Pirate Dave Silver badge

    So IBM has/had around 160 AIX developers in total? I can't decide if that seems a lot or too little.

    1. Peter Gathercole Silver badge

      Remember where AIX is in it's lifecycle.

      Most of the development on the current versions of AIX has been done over 30+ years, and as it's now mostly functionally stable, with no changes in a large part of the code base, it does not need a huge development team.

      Back in what I would imagine most people would call the heyday of AIX, the '90s and '00s, there were many hundreds of people working just in Austin on AIX development, and many more in other related developments and the Level 3, 2 and 1 product support teams in the US and around the world. When I was in the L2 team in the UK in the '90s, there were about 30 people just in that centre.

      Since about 2010, the basic AIX code has remained mostly static (which is why a lot of people complain that AIX is now quite backward compared to Linux, which is moving away from traditional UNIX at quite a pace), and the people working on AIX have basically just been working on the new features that allow things like live partition mobility and dynamic kernel patching and new Power feature, and not touching anything that doesn't need changing.

      So yes. It's not that many, but for what they're doing, it does not need to be a huge number.

  5. Fruit and Nutcase Silver badge

    New old sayings

    Old: "All roads lead to Rome"

    New: "All IBM jobs lead to India"

  6. Fruit and Nutcase Silver badge


    Advanced Interactive eXecutive

    An Ignominious eXit

    All India eXperience

    Advanced Indian eXecutive (aka Arvind Krishna CEO, IBM)

  7. Anonymous Coward
    Anonymous Coward

    I've been told the Watson Health group being redeployed in 2022 got together outside of IBM communications during their redeployment and shared information with one another, including ages, internal placement at IBM, external hiring outside of IBM, and early retirement. My contact tells me those numbers are going to be released once some legal matter are settled. IBM use to release numbers like that during layoffs, but found a legal loophole around continuing do to that, so now you will only know if you collect that info internally with other employees.

    1. Anonymous Coward
      Anonymous Coward

      There are even apparently a bunch of recordings with IBM management, HR, and leadership that were made during the Watson Health redeployment showing how IBM handled this.

  8. Liam Proven (Written by Reg staff) Silver badge

    Sad, but foreseeable

    AIX was the first UNIX I ever got to sit and play with. SMIT was a rather nice tool.

    But when I wrote -- yikes, already 8 years ago! -- that they were selling PowerPC servers that didn't support AIX, only Linux:

    ... the writing was on the wall.

    1. Anonymous Coward
      Anonymous Coward

      Re: Sad, but foreseeable

      IBM heard you and have stopped selling Power systems that do not support or run AIX!

      Also there are other developers/developments associated with AIX code, for example PowerVM VIOS,

      DS8000 storage, PowerHA (Cluster), PowerSC (Security), VM Recovery Manager, Compilers, Lab Services / Support / Cleint Engineering.

      Hence the actual number is beyond just the core AIX developer.

      In the latest AIX Strategy Whitepaper from November 2022 IBM publicly commented "AIX roadmap beyond 2035".

  9. Mockup1974 Bronze badge

    So my understanding is:

    - HP-UX: will definitely die in 2025, together with the Itanium ISA

    - AIX: might die after 2029 but nothing confirmed

    - Solaris: might die after 2034 but nothing confirmed

    - Illumos: not getting the attention it needs but being an open source project it can't truly die

    - the BSDs: from most to least active development Free > Open > Net > Dragonfly (?)

    So I wonder which UNIX will still be alive and well in, say, 2040? The only one I'd bet money on is FreeBSD...

    1. xosevp

      UNIX EOL

      - AIX EOL is Dec-2037 ( Next 7.4 ??? ) :

      - Solaris EOL is Nov-2034 (pag. #40):

      - Fujitsu SPARC products End_of_Sales is 2029, and EOL 2034:


  10. Anonymous Coward
    Anonymous Coward

    Not suprised

    IBM is run by MBA business people. When MBA business people runs a tech company, the tech degrades and falls behind. Why should IBM fund AIX when they own Red Hat? Linux development does not cost IBM nothing, so IBM have lot of Linux developers for free who improves Red Hat for free.

    MBA people is the same problem that Boeing had, with their Max 787 that crashed. The problem? It turned out that Boeing 787 had new more powerful engines. That would require a redesign of the chassis. Instead, the business people said that Boeing should instead fix the problem in software. The problem was that they did not inform the pilots of the new characteristica. Therefore the LMAX 787 planes crashed killing lot of people. If an engineer had run Boeing, they had redesigned the chassis or the the new engine.

    MBAs are good at increasing profitability but at the expense of tech and development. Another famous example is a McKinsey consultant who became the CEO of a pharmacy company. He sold all research and development, and he bought patents of medicines. That way he did not himself have to fund the research, he just reaped the benefits. Just like IBM does with Linux. So the pharmacy company valuation skyrocketed. Now several pharmacy companies are run like that. I dont know what happened to that company. BTW, Enron was also run by McKinsey MBA people.

    The problem is, Linux is not that good actually. The code quality is subpar and it has lot of design problems. Linus Torvalds famously said "Linux does not have a design, and will never have. We will rewrite all code all the time, which improves the code, and this is how evolution works. Evolution brought us humans, which proves that evolution is superior to design". There are lot of articles where Linus and other Linux kernel developers complain of the bad code quality in Linux. You see, it takes quite a long time to iron out all bugs and get the code mature. But if you rewrite all code all the time, you never get stable code in Linux.

    Some people never install Windows until the first Service Pack arrives, because the first iteration of Windows is always unstable. I have had lot of problems with Linux installations. I only run Ubuntu LTS distros to get maximum stability. But frequently some update crashes something, maybe graphics, and then I have to solve that problem which takes long time. The worst thing is when you install some new software (not LTS) which requires a new version of a library. That library update, triggers an update of all other software which use the library. And the new version of the software, use a new version of other libraries, so you must update those libraries as well. And suddenly you have left LTS and have the newest bleeding edge Linux distro. Google "FrankenDebian" for more information. The fix? Frankendebian articles say: the fix is to never install any software other than what is included in the LTS distro! So you if you want software not included in the LTS distro, you are toast, you can not install it.

    Windows is much more stable in that regard. You can install new software on older Windows distros. I never appreciated that, until I started to use Linux. Now I consider my Windows installation much more stable and less maintenance than Linux. Linux require much more maintenance than Windows 10, that is for sure 100%.

    Why has Unix lost market share? The old Unix big iron with 16-32 sockets had no competitors except Mainframes. Now even cheap servers are fast enough to handle quite big workloads. So nobody needs Big Iron anymore. Everyone has migrated to clusters instead. For Instance, Google has lot of cheap servers. When one crashes, Google just replace it with another cheap server. This way Google gets good uptime. Big Iron has superior RAS which is very expensive. But you can get as good uptime with lot of cheap servers.

    The problem is, some workloads can not run on clusters. Some workloads can only be run on a single big server, with 16 sockets. Those workloads can not be migrated to clusters. For instance, large database instances. But, there exist 8 socket cheap x86 servers today, that can handle much of that load. The hugh workloads that can only be run on a 16 socket server are far and few between. BTW, there are no large Linux servers, beyond 16 servers that are sold today. And Linux scales poorly beyond a couple of sockets. To get good scalability to 16-32 sockets, you need to go to Unix or Mainframes.

  11. TonyFCR

    IBM also stopped end of 2021 a Cooperation contract around AIX internals with Bull SAS (now ATOS) in France which started in 1992, long before AIX parts moved to India. Yes, 29 years of cooperation around AIX. ATOS still sells IBM Power machines, as Escala branding. Multi-processing was added into AIX by a Bull ~40 people team, and many many AIX tools (inside or around the kernel) were coded by this French team, which then was reduced to 30 people and then much less at the end. AIX is a great product. RIP.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like