Those of us the industry enjoy speculating what certain moves might mean -- especially acquisitions.
One hot topic is the significance of IBM's recent acquisition of Transitive. A lot of the eye-candy used the word "virtualization" (at least they didn't use "cloud"!) to position the technology.
No, I think that's a misdirection -- whether intentional or not.
Transitive's claim to fame is translating instruction sets and providing lightweight runtime environments between processor families.
For example, when Apple went Intel instead of PowerPC, Transitive's technology was a key part of keeping their customers happy. I don't know how many people actually used the stuff, but I think it was a key part of the optics, anyway.
Might another processor transition be coming up for IBM?
Given all the news about Intel's (and AMD's) roadmaps, how much enthusiasm is left for keeping PowerPC competitive? The stakes just keep going up, and up, and up.
Now, I don't expect them to admit this publicly, but inquiring minds just have to ask the question, don't we?
Let's see, there's the AIX pSeries family (a healthy part of the UNIX market), the iSeries (AS/400 to all you historical types), and even a variant of Linux for pSeries that never quite found its stride.
Imagine alll that on x64. And all that now a candidate for VMware-style virtualization.
Maybe in the future, we're only going to see two major operating systems in larger data centers: zOS for all the big stuff, and VMware containerizing just about everything else.
I, for one, am convinved that the "big UNIX machine" market is going to face serious pressure from VMware at scale. And recent processor announcements have done much to encourage me in this line of thinking.
Maybe I"m not the only one who came to this conclusion? This should be interesting ...

I'm with your line of thinking on IBM POWER - it just doesn't make sense to put energy into high-end microprocessors anymore - we've moved to an x86/x64 monoculture.
As for operating systems, I certainly agree that zOS isn't going anywhere. Windows is solid in the data center, and other UNIXes are losing out to Linux.
VMware is solid, too, with an impressive strategy to become the standard platform for non-mainframe systems. But what will it host? Windows and Linux guest application servers and specialized appliances of all stripes.
So we get variation within VMware, but variation of open systems hardware is a thing of the past.
Posted by: Stephen Foskett | November 19, 2008 at 02:00 PM
Interesting. HP have Aries for RISC and had Dynamo (years old now), and there have been various dynamic binary translators around for some time. There was also a static binary translator called FX!32 for the Alpha to help in migrations after HP shot it in the head way back after aquiring Compaq.
If the PowerPC dies, it'll be for other reasons, and I reckon this isn't why IBM have purchased Transitive. Static translation as with FX!32 would be the preferred route, but Transitive's product appears to be a dynamic translator from what I can work out.
I'd suggest that the point here *is* virtualisation, and no misdirection and speculation is needed.
It's not about operating systems either. It's about instruction sets (although, granted, there are OS hooks required). It's the kind of technology that allows you to dynamically move workloads from Sparc to x86 to PowerPC. VMotion that's chip independent, if you like. I seem to remember VMware doing some research in that area, though I can't find it right now.
Posted by: Alex McDonald | November 19, 2008 at 02:00 PM
If that happens....game over for a number of vendors. Tell you what tho' Chuck, IBM's virtualisation in pSeries and obviously zOS is still the gold standard. VMWare may be getting there but there's still that tricky bit of the mountain to climb.
Posted by: Martin G | November 19, 2008 at 02:29 PM
Does this enable IBM to create a x86 Virtual-Appliance (on vmware or xen) of a powerpc application ?
Posted by: uday | November 19, 2008 at 04:11 PM
I don't believe for one minute that IBM purchased the product to eat its own. P-series running Power architecture will continue for a while yet, due mainly to the fact that there are apps that only run on P-series and I-series and secondly that there are customers willing to drop large six-figure sums for each server, despite the preponderance of evidence that they can drive costs out of their business by deploying standards-based x86 reference architectures.
No, IBM's buying this because it's a good product set - it works - and secondly they see the opportunity to further tighten the thumbscrew on the other proprietary Unix vendors - HP (HP_UX/Itanium) and Sun (Sparc/Solaris).
If the statistics are to be believed and the Unix space is reducing, IBM's looking to get a greater share of this declining space.
I see a virtualization angle, but I don't see a direct VMware angle, Chuck.
Geoff @ Dell.
Posted by: Geoff Mitchell | November 19, 2008 at 06:09 PM
I agree with Alex - IBM probably didn't acquire Transitive to EOL POWER. Its much more likely that they did it to increase the leverage of POWER.
Posted by: marc farley | November 19, 2008 at 06:55 PM
Wow, great discussion guys, so thanks!
Just to be clear, I don't think IBM is in love with VMware (quite the opposite). I think they're looking for a clean way to get their loyal base from Power to x64.
One of the side effects of doing so, though, is to widen the opportunity base for VMware.
For those of you who think IBM's intent is flowing in the other direction (e.g. x64 to Power), you've got a point, it's just unrealistic enough to make sense to some product guy at IBM. Doomed, though, if that's the case.
Martin is right, virtualization is extremely mature on zOS and AIX. No argument there. But I tapped my resources looking for a PowerPC processor roadmap, and there wasn't much to be found out there. In a world where other CPU vendors are roadmapping out until 2012 and beyond, that's a bit indicative, in my book.
Now, x64 binaries running on zOS? I'd pay to see that.
-- Chuck
Posted by: Chuck Hollis | November 20, 2008 at 03:08 AM
On reflection and reading a bit further about this, I think I understand what IBM are doing.
There's a confusion in this piece and some of the comments between hardware (chipset) and OS. Most (all?) binary translators are tied to Unix, so they can, for example, translate SPARC instructions sets on UNIX compiled binaries to x86 instructions under UNIX.
Z series hardware can run UNIX natively. So an x86 UNIX binary could be dynamically translated to run on the Z running UNIX; it's not rocket science (close, but not quite). I'll take your money now :-)
As Martin points out, virtualisation is something that IBM knows a lot about. They've over 4 decades of experience with it. This is definitely a technology that can help with doing this across hardware platforms.
But I don't believe that IBM are VMware unfriendly, or Xen or Hyper-V for that matter. In this case, they don't care.
The aquisition of Transitive is not about x86 to PowerPC or Z; it's about running SPARC application binaries for which you don't have the source. IBM wants to run your SPARC UNIX apps on their UNIX tin. Right now, I reckon it's Sun they're gunning for.
Posted by: Alex McDonald | November 20, 2008 at 06:16 AM
OK, now that makes sense to me as well.
Gunning for Sun seems to be a reasonable tactic these days, but given that so much that I see in the Sparc environment is either open source or already on AIX, one has to wonder if they'd pay all that money for a binary translator
Posted by: Chuck Hollis | November 20, 2008 at 11:20 AM
Lets take the PowerPC architecture; how many times today will you interact with something which uses PowerPC in some form or another? IBM do make an awful lot of money out of Power and I'm not expecting them to dump it any time soon.
BTW, I saw a roadmap a few years ago taking it beyond 2012; now, that might have changed as I've not had anything to do with pSeries for a couple of years.
Marc is certainly right, they are going after the Sparc base very aggressively. I've seen some very attractive prices thrown around; not attractive enough to convince people to change from the cult of Solaris to the cult of AIX yet. If they have a translator which allows a user to take their binaries across and run; it could be a very powerful sales tool.
'Hey Mr Customer, we can show your bespoke application running on our tin....and we didn't even need to recompile it'
That could be a very powerful pitch.
Posted by: Martin G | November 21, 2008 at 02:47 AM
On the question of if this will be used to allow people to GO TO powerpc or GO FROM powerpc, keep this in mind.
Transitive was the people who were instrumental in writing the part of MacOS that Apple used to allow people to seamlessly run programs compiled for their old platform (PowerPC) on their new Intel based systems.
So the best thing they are known for is for allowing POWERPC programs to run ON INTEL, not the other way around.
They are also known for allowing solaris sparc based software to run on Intel...again....foreign platform running on Intel.
Those are stronger clues as to a potential direction than about anything else you can come up with I think.
Posted by: Han Solo | November 21, 2008 at 09:50 AM
(1) I think putting Transitive in the virutalization category is wrong.
(2) I think there are several things at play here including SUN, CIO agenda and financial meltdown...
I posted a blog about this at http://soa-biz.blogspot.com/
Posted by: babak hosseinzadeh | November 21, 2008 at 12:54 PM