Two of IBM's oldest and most popular operating systems for its Power-based servers are being put out to pasture after years of service. Last week, IBM said that it would be offering service extension on AIX 5.3, the operating system that was announced way back in July 2004 concurrent with Power5-based System p5 and i5 iron. …

COMMENTS

House rules Send corrections

This topic is closed for new posts.

Friday 17th February 2012 18:52 GMT Anonymous Coward

Related to ongoing contracts that are late and run on AIX 5.3 perhaps

IBM tenders on a contract, uses latest and greatest, fails to deliver a 2 year project a bit late and all that lovely tech is out of date.

That all said it would be somewhat pushy to assume that everybody updates OS every time you release it. If the curent one works fine and the kit is due to be replaced/upgraded in a 5 year cycle; let alone that they sync up with IBM's release cycles, then unless it's LPAR'd up onto a newer box it will just sit were it is, working, doing it's job, working fine. Also upgrading any OS version, nomatter how smoth will always have gotcha's (thining MQ queue storage few years back myself) with the software that runs upon them.

Another view of all this could be that IBM is very good at playing poker and only when enough call there bluff do they play the cards they were prepared to play from the start. But at least gain some panic early upgrades and most people who get a AIX box will in general just apply PTF's (permenant tempory fix's(sic)) were needed and like everybody know's, if it works and works well without issue then who are you to introduce issues.

Bottom line those who did not but the latest kit on day one of its release will still get tehre 5 years support and accountants will sleep at night, curse you IBM, curse you.

1 1
1. Friday 17th February 2012 20:02 GMT pPPPP
  
  Re: Related to ongoing contracts that are late and run on AIX 5.3 perhaps
  
  "PTF's (permenant tempory fix's(sic))"
  
  Why "sic"? I've never seen IBM write PTFs as anything other than "Program Temporary Fixes".
  
  Who are you quoting? And why can't they spell?
  
  Yes, people are quite happy to leave well alone. I have been forced to use Windows 7 on my laptop and I can't stand it. XP was fine.
  
  So do you expect your OS vendor to support your OS ad infinitum? Do you expect them to continue educating people and paying them to wait for your phone call? What do you do in your job? Same thing you did 5 years ago? Are you happy to do so and keep getting paid the same and have no prospect of career progression? Or would you be happy to move into a new job just to learn an OS that's 5 years old? That would look great on the CV.
  
  Point is, to support something you need to have people who know it in depth, and be good at it. And you need to pay them and give them careers, or they'll move on.
  
  So you're welcome to use your old OS for as long as you want (and it's licensed). If you want support, you're going to have to pay for it or do without.
  
  At least with virtualisation you now have the option to consolidate OSes and get some way of running it on new hardware.
  
  1 1
  1. Saturday 18th February 2012 17:03 GMT Rippy
    
    Re: Re: Related to ongoing contracts that are late and run on AIX 5.3 perhaps
    
    PXG may be excessively modern in his liberal use of apostrophes, but he is absolutely correct in pointing out the oxymoron in Program Temporary Fix: they're as temporary as income tax. The original idea when the terminology was coined back in the last century was that they were temporary patches; now they're universally acknowledged as lasting until hell freezes over.
    
    1 0
  2. Saturday 18th February 2012 19:16 GMT Jim in Hayward
    
    Re: Re: Related to ongoing contracts that are late and run on AIX 5.3 perhaps
    
    Well, I don't know about you, but I pride myself on having actual working knowledge of 80% of every OS on the planet. It does make it difficult but I think that IBM did a great thing with their OS being able to virtualize the older OS. I can't keep 10 servers around my home office!
    
    I don't think anyone is complaining about IBM's approach. It is good, fair to those that just will not budge, and a good incentive to move to new hardware and OS.
    
    This is how Microsoft should work. Sadly, MS will be dead in 12 years when the hanger-oners (lol) finally give up.
    
    0 0
2. Friday 17th February 2012 22:00 GMT Peter Gathercole
  
  Re: Related to ongoing contracts that are late and run on AIX 5.3 perhaps
  
  PTF stands for Program Temporary Fix.
  
  AIX 5.3 was actually released in 2004, and the product lifetime cycle published on Fix Central has indicated that TL12 would go out-of-general support later this year as early as this time last year.
  
  This means that AIX 5.3 will have had a lifetime of over seven years, and if you take into account extended support, will be more like 9 years.
  
  TPM's article has several errors in it to do with dates and functionality of earlier versions of AIX (like LPARs being introduced in AIX 5.1). I've sent a correction, so we will have to see whether the article is fixed.
  
  0 0
  1. Friday 17th February 2012 22:26 GMT tpm
    
    Re: Re: Related to ongoing contracts that are late and run on AIX 5.3 perhaps
    
    Just dyslexia on the date
    
    The IBM HW and SW manual has dates "2004/07/13" and I flipped the month and year bits because, as an American, I wanted data in the "07/13/2004" form.
    
    0 0
Friday 17th February 2012 20:18 GMT regadpellagru

"On the proprietary side of the IBM midrange house – as if the Unixes were not as proprietary as z/OS, OS/400, OpenVMS, and Windows, but this is the language we inherited from the 1980s ..."

Ah ah, kuddo to the author for this good one, indeed very true. Decades after the last common trunk of Unixes (SYS V R3, if memory is not totally gone), what's left of the "standard", as a common denominator between them would probably fit on one A4 sheet of C source code !

Maybe something linked to "/" as a directory/file separator ....

Industry loves to split up integrated things ...

0 0
1. Monday 20th February 2012 12:46 GMT Anonymous Coward
  
  Huh?
  
  So you've never heard of XPG4, POSIX:2008 or SUSv4?
  
  0 0
Friday 17th February 2012 20:23 GMT Anonymous Coward

Less marketing

What is better for long term revenue, suppliers, vendors and users is the time based versioning, like Ubuntu LTS: You've got several years to prepare your applications, middleware and hardware for the migration to the latest version. And you know when the next release is coming.

Frankly, if you don't listen to IBM's marketing guys, AIX 6.1 and 7.1 could have been numbered 5.4 and 5.5 respectively. But there were Power 6 and a Power 7 processor releases inbetween, and a split in Express and Enterprise edition, it all generates revenue.

Which leaves a question for Timothy Prickett Morgan: is there a comparison between Linux and Unix, as to how many innovations are between two (LTS) Linux versions and two (Solaris, AIX) Unix versions?

From what I've seen, since 2008 Unix is playing catchup to Linux, when it comes to innovation.

0 0
1. Friday 17th February 2012 23:35 GMT Peter Gathercole
  
  Version numbering and catchup
  
  I agree with your comments on version numbering. There is not that difference in the outward appearance of AIX 5.3 and AIX 7.1. Under the covers, there have been quite a lot of changes, including dynamic partition migration, more control of WPAR isolation, support of more logical CPUs, USB storage support and up to 4 threads per core.
  
  A lot of the changes are targeted to the very largest systems, but this is not really a problem because what features that you need are not in AIX already? I'm not including things like Gnome and KDE, because they are not really part of the OS. DBUS maybe? Remember that AIX has never really been a desktop OS, and much of that Linux innovation has been in things that are really applicable to personal systems rather than servers.
  
  I don't rate your comparison of AIX and Ubuntu LTS. In the same time I have been working with AIX 5.3, I've gone from Ubuntu Dapper Drake (6.06), Hardy Hedgehog (8.04), and Lucid Lynx (10.04).
  
  IBM have always clear about the lifetime of it's OS products. End of marketing is always announced at least a year before it is actually withdrawn (and normally soon after the +2 version is announced), and there is normally at least a year of support from End-of-Marketing to End-of-Support, and then there is always extended support for customers prepared to pay. And after that, the mature AIX product (which after so long in support is likely to have had all of the serious problems fixed) will have the fixes available on fixcentral for a couple of years more.
  
  0 0
  1. Sunday 19th February 2012 13:50 GMT Anonymous Coward
    
    Re: Version numbering and catchup
    
    What you are listing here what has happened under the hood are additional features which IBM had to implement because the competition came up with these ideas. Of course, if that Power7 processor now has 4 SMT threads these need to be mapped to 4 logical processors in AIX. This functionality was added to AIX 6.1 with TL05, so that AC was right in saying version numbering has a lot to do with marketing. And never mind the 8 threads of SPARC T3 and T4.
    
    Have you looked at AIX commands? Many of these are ksh and perl scripts, and they have comments, like #Fixed as part of APAR IX123... this tells you the last time this script - ahem - AIX command was updated. Have you worked with AIX L3 support? When you find the workaround and fix faster than they do, and you start writing your own tools because simple commands like lsvg and restvg are simply not up to enterprise standards anymore then you realize you have crossed the line, you have started writing on your own OS. You might as well start your own Linux distro.
    
    0 1
2. Sunday 19th February 2012 05:47 GMT Anonymous Coward
  
  Re: Less marketing
  
  Unix is playing catchup? Linux is STILL trying to copy ZFS and DTRACE. Their network stack can't touch crossbow, and they haven't even attempted to create a unified block project like COMSTAR.
  
  So no, Unix isn't playing catchup, it's continuing to lead the pack. Hell, has a modern linux OS managed to create static device entries yet 20 years later?
  
  1 0
3. Sunday 19th February 2012 16:13 GMT Anonymous Coward
  
  Re: Less marketing
  
  Ridiculous. Linux is a Unix wannabe. And still tries to copy Unix, at least Solaris with zfs, dtrace, smf, crossbow, etc. I dont know about AIX though. Apparently AIX development pace has slowed down, only IBM knows why. So it could be true that AIX is catching up on Linux, but Linux is still trying to copy functionality that Solaris has today. BtrFS is just a huge zfs ripoff. Systemtap is ripoff of Dtrace. AIX probevue is a Dtrace ripoff. Linux is trying to copy solaris zones, as IBM already has copied it (WPAR). Linux is also copying Crossbow now, forgot the Linux name. Something with... openstack or so. Aix is catching up on Linux. Linux is catching up on Solaris.
  
  0 2
  1. Wednesday 22nd February 2012 12:16 GMT Anonymous Coward
    
    Re: Re: Less marketing
    
    Yes, Linux is Unix-like. But if you see the amount of research and innovation going into it, it's easy to see that Linux is taking away the innovation crown. But don't take my word for it, just crawl a bit through Wikipedia, and you will see that UNIX has been marginalized in HPC/Supercomputing. Just have a look at Japan's K computer and read up on the optimized parallel file system based on Lustre and the six-dimensional torus network.
    
    Linux is like a Lego box - you build it to get the perfect solution for you. So should be the infrastructure and application solution in your company.
    
    0 0
Sunday 19th February 2012 08:45 GMT T J

Can they follow through and do it to themselves?

Can we put IBM out to pasture too now? Or even better just put it down? And all its employees and products with it?

0 0
Tuesday 21st February 2012 20:47 GMT Michael Wojcik

1978?

"This approach gave the AS/400 the illusion of application compatibility, but every now and then – in this case, three times in the history of the line since 1978 – the underlying hardware is so different that the abstraction layer in microcode also has to be changed to make use of the new iron."

Is "1978" there a typo for "1988"?

The AS/400 "line" did not originate in 1978; it was released a decade later, in 1988, as the conclusion of the "Silverlake" project to develop a follow-on line to the System/38. And while the '400 inherited a lot from the S/38, and somewhat less from the S/36 (which was a sibling of the S/38, both succeeding the S/34), plus some other ideas left over from Future Systems, the S/38 (released 1979) was not an AS/400.

Many S/38 apps could be run under OS/400, but in a special emulation environment. They weren't converted to run as normal OS/400 program objects. (In the AS/400's CISC-to-RISC transition, program objects were converted, as the article says. That could actually be done while installing program objects from external media; we used to ship AS/400 software that could be installed on CISC or RISC machines.)

0 0