IBM Unix hit by false Falco
But Caldera left holding the baby
Since IBM has been referring to its Monterey development project in the past tense for a few weeks now, we were surprised to see that this stole a few headlines in the trades last week.
Whether Monterey is alive or dead depends on whether you want to distinguish marketing news from technical news - there's quite an important difference here. So let's have a look.
Monterey is the code name for the joint IBM/SCO project to bring elements of AIX, UnixWare and Dynix/ptx together on both IBM's Power RISC chips and Intel's IA-64.
There was never any doubt about who held the boss hand in this deal. Big Blue needed a Unix on Intel the most badly, in case Itanic should unaccountably become a major player. Meanwhile SCO and Sequent - who had the high and low ends of the Unix on Intel business between them - could have got to IA-64 under their own steam, but they both faced big R&D spends to get there, so pooling their resources made perfect sense.
IBM's manager for "deep computing" David Turek was in town last week and told The Register:"Monterey was a project to make AIX a fully 64bit operating system, and make it run on POWER and IA-64 ... and we have done that," he said. "It is out in beta, people are doing applications on it, some OEM partners are working on it - and that is the evolution of Project Monterey."
Which is fair enough. We never expected IBM to drop the AIX brand that it's spent ten years investing in, and we suspect that even in SCO founder Doug Michels' wildest fantasies (don't go there - ed.) AIX on Itanic would still be branded as AIX on Itanic, and not as some Super-UnixWare. Right from the start IBM folk were telling us that this would be the case. So how much of Monterey has died? Nothing, as far as we call tell. Code names come and go, and Caldera (UnixWare's lucky new owners) get to keep an IA-64 port of UnixWare that stoic SCO insiders insist they'd rather not have.
So say farewell to one marketing wheeze - and say hello to another.
IBM's Turek tells us that the 'L' in AIX 5L stands for - you've guessed it, Linux. Purely as an OS - discounting any applications support - the Linux IA-64 port will be out of the door at about the same time as the "Monterey" UnixWare, thanks in large part to the investment from Intel, VA Linux, SGI, Hewlett Packard and er... IBM.
So where does this leave UnixWare? Well, let's not forget that the market has steadily been collapsing from under the venerable OS - with revenues in the most recent two quarters half of their previous levels. And that's not through some technical oversight: UnixWare is far and away the most scalable and capable server OS for PCs. It's just that the momentum of Windows 2000 one one side, and Linux/BSD on the other has absorbed much of the growth in the server market. The real problem for UnixWare is that its new owner Caldera doesn't have much of a vested interest in advancing the 64bit port. Quite the opposite, in fact: if it's going to remain both a leading free software company and remain a leading server OS vendor, then it's going to have to put it's eggs in the Linux-on-IA64 basket one day. That day may not come just yet - Intel's next-generation 32bit P7 processors already look very competitive - and Intel expects to five to seven years life out of derivatives from these cores. And of course, AMD's x86-64 ought to find a niche, at least for a while, as it offers an incremental path for customers who really need the 64bit addressing.
But while all this goes on in the coming years, those Caldera R&D dollars spent on maintaining the UnixWare-64 port being to look increasingly dubious, especially against a background of Linux-on-IA64 gaining the scalability, reliability and performance of UnixWare through various free software efforts. And some of these, like ReiserFS (a high-performance journaled file system) and FailSafe (HA clusters), are already here. You can be sure that rivals SuSE and Red Hat and TurboLinux will be focussing their efforts on IA-64, so those development bucks will have to be justified - and justified convincingly.
So unless IA-64 turns out to be a complete, long-term dud, and Intel conjures up some kind of "official" 64bit CISC (this is the subject of heated speculation, by the way) it looks like Caldera has spent a lot of money on a pup.
It's almost ten years since Borland bought Ashton Tate, and much was made at the time of how it could thus become a one-stop shop - a software powerhouse to threaten Microsoft. A few dissenters at the time wondered why it had paid so much for a product - dBase -that overlapped with its own portfolio, and Borland spun ever more creative tales of how it intended either to merge the products, or alternatively target each database at different niches. At one stage it even tried to pretend they were one and the same thing. But dBase soon began to look like a legacy product without a long-term future. The ill-judged merger sank Borland's grand ambitions, and eventually cost Borland founder Philippe Kahn his job.
It's far, far too early to judge if history is repeating itself, but many of the ingredients appear to be there for the unwary. ®
IBM claims AIX 5L is proof of Monterey concept
IDC pricks SCO-Caldera bubble
Double-spinning Caldera faces open source backlash
Linux beating Win64 to Itanium punch
Oracle adds support for IBM-SCO IA-64 Monterey