AIX On the Desktop Is Getting the Boot 366
flnca writes "Today, I was playing with the thought again to purchase an AIX workstation one day when I can afford them, and I was surprised to see that IBM is going to give its IntelliStation POWER Series workstations the boot in January '09. A black day for AIX on the desktop. I really wonder what's the problem there, warehouse costs? IBM has a history of burying its best stuff (like OS/2 for instance). Some years ago, I enjoyed hacking away on an RS/6000 workstation running AIX 4.2, and it was a pure joy. Not only the kernel, but also the admin tools, like smit and smitty. Their blade-centric solution uses Windows as a client for workstation application. This truly sounds like IBM wants AIX only for servers anymore. I'm not amused. Although, eXceed on Windows with an XDCMP server running on AIX might also be a viable solution ... whatever. But it can't beat a native POWER box sitting on your desk, that's for sure."
It's not just you (Score:5, Informative)
From testing OS & firmware upgrades to just being a great desktop platform, it's proven to be very valuable.
- Tony
Re:No, (Score:5, Informative)
It was submitted by someone in Germany (so English is most likely a second or third language). It was edited by... well it wasn't edited.
Go ahead and suck it up. (Score:3, Informative)
AIX is horrendous. I mean, truly horrible.
Smitty - though it has its uses - is the nastiest piece of manure ever to disgrace an SSH window. Everything even remotely UNXy IBM makes is, IMHO, totally over-priced.
AIX hardware is over-priced, under-powered and totally uninteresting. I have machines running Linux on Opteron right here and they simply out-perform AIX machines (including a 12 CPU Power6 P570 AIX 5.3) at least 10 times.
And don't get me started on the stability of AIX vs Linux or BSD, please. I have software here that can make any AIX machine cry and call for mommy, when most Linux distributions just suck it up and carry on.
AIX machines are essentially dull ultra-expensive big iron. Most programmers I work with would rather have a small machine with Red Hat and tons of GNU goodness on it than a huge AIX beast.
And just in case you are wondering: yes, I do administer UNIX machines for a living. Just check my Slashdot journal, and you'll get a ton of information on AIX, Solaris and so on and so forth.
This being said, I'll take AIX over Windows any day. And either Slackware or OpenBSD over everything else.
The march towards Linux (Score:4, Informative)
Over time, all the cool features from proprietary UNIX versions are getting ported to Linux, either directly or by being re-implemented. As Linux becomes more and more acceptable as a replacement, expect to see proprietary UNIX versions start to go away.
If IBM hires a person to work on Linux, that work helps IBM across pretty much their whole product line. If IBM hires a person to work on AIX, that work has much less value now, and will have even less and less value over time as Linux gathers up more of the market. Also, as Linux keeps getting better, it would take more and more work to add similar features to AIX, to try to keep up. Eventually, IBM is going to stop paying for work on AIX at all; they will end-of-life AIX, and just sell Linux.
I don't know for sure about SMIT but Linux does have LVM and various tools to manage it. AIX gurus, how ready is Linux to replace AIX now?
And, are desktop POWER machines going to be available with Linux?
steveha
Warehousing Costs (Score:4, Informative)
Warehousing costs are an indicator not a base cause. If you have 1000 units sitting in a warehouse for six months depreciating, it's because no one's buying them. Which means you're losing money from a failed projection. Something this seemingly slow moving would likely need a different supply chain, say direct from manufacture, JIT. Also, the margins on such might just not be there. Hardly worth the effort since IBM is not a non-profit.
Re:Don't be silly (Score:5, Informative)
Except that the 5 GHz CPU is a POWER 5 processor (if I am right), that beats the living shit out of AMD or intel, when it comes to computational power per clock cycle. ;)
I recommend Xming instead of Exceed for X (Score:5, Informative)
If you're just need an X server on Windows to connect to your *nix box, I suggest using Xming [sourceforge.net]. It's free, lightweight, easy to configure, and one can quickly setup shortcuts to connect to a specific server and run a program. It's also very useful for getting around a content filter if you can access your own *nix server from the internet.
I don't have any affiliation with Colin Harrison, however I've used other X servers on Windows before and this has been the best. Here's my experience with different X servers:
Exceed - Bloated, expensive, extra licensing fee for doing X11 over SSH, unstable copy and paste (in the past versions I used)
ReflectionX - A bit bloated, expensive, funky interface
Cygwin* - Too many unneeded apps included for just an X server, FREE, difficult to configure if you're not familiar with it
Xming - Light weight, FREE, quick install, can use PuTTY's plink to do configure free X11 forwarding over SSH, copy and paste works, it just works
*In regards to Cygwin, I understand that it is more than just an X server, however it has been recommended a number of times to me as a solution for a free X server on Windows
Re:It's a true desaster. (Score:4, Informative)
Re:It's not just you (Score:5, Informative)
I used to be an AIX administrator.
There's not a lot of benefit to having an AIX box on your desk (though I did), other than it being the same as the systems you're administering.
(The following is my personal opinion - fanboys of other operating systems need not respond; I'm sure your OS of choice is just peachy too)
Yes, AIX is more stable and I prefer the management tools and interfaces to other Unix-like operating systems. As such, having it on my desktop was preferable to a Linux system because I was more familiar with the tools and they were the same as the machines I was administering all day long.
If I was running Linux systems for a living, I'd have a Linux box on my desk for the same reasons.
There are some advantages to writing/testing your code/scripts/etc on your local machine before pushing it out to a development/production system. While in theory ksh/bash/csh/etc should be the same on every system, we all know there are quirks to the implementations that cause issues.
So yes, there are some benefits to AIX on the desktop as an administrator.
Finally, there are some shops (a few military contractors I'm familiar with) that use AIX on the desktop for their engineers because the specialized applications they use only support AIX - usually graphic design hooked into large AIX systems on the backend for modeling/redering cycles.
Re:Go ahead and suck it up. (Score:3, Informative)
I agree with some of your comments.
SMITTY is ugly, but I think it's a good tool. The best feature is that it constructs the command line commands rather than trying to modify configuration files or re-write the tools. This means that anything that you can do via smitty can be easily scripted even if you don't have much AIX experience.
For some workloads Linux will kick the pants of AIX. For others, especially those that require high throughput, the story is different. AIX on pSeries can move massive amounts of data, more so than a similarly configured PC based server.
AIX has some awesome disk tools. I use Linux on a daily basis, but the Linux tools are not yet at the same level as AIX. The current state of LVM is about where Veritas was a couple years ago. This is still enterprise quality (and free, dammit), but generally not as easy to use as AIX.
And yes, I also administer AIX, but have been running Linux in production for more than a decade.
Re:It's a true desaster. (Score:3, Informative)
Re:Go ahead and suck it up. (Score:2, Informative)
Once you know your way around the pitfalls, it's OK-ish to run, administer and use. But, given the cost of the OS and the hardware, why bother?
Re:They made that? (Score:3, Informative)
AIX was available on x86. Years ago (around 1991?) IBM gave my company an AIX package for PS/2 hardware. Gobs of diskettes, pounds of printed docs. It seemed clunky and we never did much with it, but it was there. Of course the PS/2 Model 80 (i386) it was installed on was a slothful piece of crap.
Re:Go ahead and suck it up. (Score:3, Informative)
I used it ages ago and remember thinking it was the most broken software with a version number past 3 I had ever seen. Non-standard and quirky, too.
Re:It's a true desaster. (Score:3, Informative)
CDE is still standard on Solaris (you can choose between CDE and GNOME at install time), which runs on SPARC and x86 systems.
True, but CDE is no longer supported on OpenSolaris [opensolaris.org]; which is a much better choice for desktop user than Solaris 10. The upcoming OpenSolaris 2008.11 version, and update to OpenSolaris 2008.05 has many more improvements that make it a viable alternative to GNU/Linux on the desktop or laptop.
Re:A Huge Blow (Score:4, Informative)
Where have I been the past decade? In some of the most influential aerospace and aircraft design firms in the world, actually.
Yes, the video performance of the Quadros is undeniably fantastic. I even use retired Quadros in my home machines. Not every engineer needs a POWER based machine, of course. Many engineers could do just fine with a Windows box. But, serious CATIA work, meshing, and analysis were impossible to do on Windows machines; they simply couldn't touch the AIX boxes when you needed to run something that would require more than 3GB of RAM. Right now, Cessna Aircraft is still using POWER based machines for a very large portion of their CATIA work, although they were starting to transition to Windows boxes. Everyone there who supports CATIA and ENOVIA has a POWER workstation.
Now that Windows and the x86 CPU family has gotten with the program, they have barged their way into the engineering computing world. Cheap multicore processors and cheap operating system licensing makes the decision today to use Windows PCs a no-brainer. Now that Windows boxes can do what the AIX boxes have been able to do for a long time, the cheaper Windows boxes are finding their use on engineering desktops, and software developers are writing for Windows. But the point is that this is a very recent development. As of 2005, it was *impossible* for me to do the work I needed to do on a Windows box; the technology (hardware and software) *did not exist.* Price/performance is irrelevant if performance is zero; if a box can't do the work you need it to do, it doesn't matter that it was cheaper than some other box.
Re:No, (Score:3, Informative)
A more clear example to illustrate this would be "He is talking" would directly translate into "Esta hablando".
"Esta" being the verb "estar" (to be) conjugated with the third person pronoun "el" (he). That's why the construction "el esta hablando" is redundant, the "el" is only needed for emphasis, which would be something like: He, he is talking... but "esta hablando" would translate to: "he is talking", and not: "is talking", even thought there is no explicit pronoun present.
The "tool" has experience in assembler (Score:5, Informative)
It was a hardware limitation, you ignorant tool.
Sorry to disappoint you : I'm not a tool.
In fact, I happen to have quite some experience programming assembler for x86.
There's no such thing as a 640Ko hardware limitation. That number is completely arbitrary. Pulled out of Bill Gates' ass.
The 8088 and 8086 chip have 20 address lines. Meaning : 2^20 addressable byte or 1 MiB memory limit. The limit is there at 1 MiB.
When designing the memory layout, they had to reserve some address range to be used for stuff other than memory (BIOS, address range used by hardware, etc.)
You have a couple of actual limits imposed by the 8088/8086 chips :
- Memory is up to 1MiB
- As small portion at the begin of the memory is used for the interrupt table.
- The last bytes before the 1MiB are where the processor starts when turned on and contain instruction to jump to the BIOS it self.
These are the only fixed addresses
The split between physical memory and mapped address space could be placed anywhere.
640k was just chosen because :
- it's ten time the 64k addressable by previous machines
- it's the first segment beginning with a letter in hexadecimal. memory is in segments 0000 to 9000, reserved are in segments A000 (color graphics) to F000 (BIOS)
If the addresses hadn't been fixed in advance and/or the reserved space had been place in the begin of the address space like on most home microcomputers, the address space left for memory would have been continuous. Yielding to more free addresse for more "main memory" (the upper 384 are a huge waste of space - as proof see all the TSR programs that existed to try to "loadhi" and cram more software in that "UMB" memory range). A continuous memory scheme would probably have helped a more easy transition scheme to processors with bigger address space.