The AIX and pains of 64-bit computing

I imagine that, by now, some folks at Caldera have noticed the fact that their company is involved with no less than six operating systems.

First there are Caldera's present Linux-based products, eDesktop and eServer.

Linux IA-64 is the partnership formed to ensure that Linux gets the most from the IA-64 (Intel's 64-bit CPU architecture). And OpenServer and UnixWare are the two long-running Unix variants the company obtained when it bought SCO.

And then there's AIX L.

"But wait," I hear you ask. "Isn't AIX the name of IBM's proprietary Unix variant?"

Well, yes it is. Thank you for asking. And herewith we find what could rightly be known as The Operating System Formerly Known as Monterey.

Just about two years ago, SCO, IBM and a bunch of other well-meaning Unix companies decided that it would be better to combine efforts than to continue to produce similar but competing Unix platforms. Code-named Monterey, the project was designed to enhance UnixWare's 32-bit capabilities while working to have a 64-bit Unix ready in time for the release of the Intel CPU, then called Merced.

It's an understatement to say that much has changed since then, most of it having to do with this Linux thing. IBM has jumped on the Linux bandwagon with both feet, both hands and a bunch of other body parts. SCO went through a love-hate-love evolution with Linux until it was recently purchased by Caldera, a Linux specialist a fraction of SCO's size that now has the task of juggling these six different operating systems.

Most interesting has been Monterey's evolution away from a project that was supposed to bring together the best of all vendors. While Monterey was originally hyped as a merged product, UnixWare administrators looking to migrate to AIX L are in for a surprise, while existing AIX installations won't have much of a problem at all. Monterey has appeared to become simply the next release of IBM's Unix, with an IA-64 port, some Linux capabilities, enhanced multi-processing and multi-platform UDI driver technology.

SCO's Monterey product manager, Scott Allen, has no problem with the IBM-specific naming or the project's general evolution into a hierarchy of companies (with IBM firmly entrenched at the top) rather than a cooperative partnership.

"AIX has broad recognition in the enterprise," Allen said, "and it's very important for the combined SCO and Caldera to have a presence in that market." Allen said that AIX was more advanced than UnixWare in features that are necessary to this highest peak of the market, and it has more available applications.

Furthermore, Allen said that the AIX recognition will be helpful when approaching the first potential market for IA-64-based systems. "Our market research shows the early adopters will be RISC users who want to improve their price-performance." Scott said IBM, SCO and others were aiming squarely at Sun's share of the large systems market.

According to Miles Barel, program director, AIX marketing at IBM, it's all going according to plan. "We have done exactly what we said from the beginning," he said. According to Barel, the new AIX is incorporating two UnixWare features -- the /proc file system and the System V printing system.

One of my own questions about all this was how SCO, a company committed to hardware vendor neutrality, was going to sell and hype an OS whose name was so tied to one specific player. Would companies such as Compaq and Dell, who have been pretty closely tied to SCO, be enthusiastic about selling an OS made and named by a direct competitor? According to IBM's Barel, "We decided to retain the name AIX on the advice of our customers." Well, I'd expect that from existing IBM customers. But wouldn't existing UnixWare customers and hardware partners get scared off?

Allen didn't think this would be a problem, because AIX L will give these OEMs a way to compete against RISC vendors in a way they haven't been able to do before. "Are they really excited about the product being named AIX? Of course not," he said. "Will the name prevent them from promoting it? No. This is mainly about making money and the OEMs know they can do well with AIX."

Not everyone within SCO is as enthusiastic. One who (understandably) requested anonymity volunteered this:

"SCO worker bees have been screaming about these horrors for most of the two years this 'partnership' has been in place. We're mystified as to how this relationship got this far out of control. From what I can tell, we exchanged several person-decades for the privilege of being an AIX VAR."

Strong words, but ironically the commandeering of Monterey by IBM may suit Caldera's interests better than those of the pre-Caldera SCO. Having IBM maintain AIX L at the very highest end -- in the thin air of 32-CPU systems and huge installations, which Linux won't be ready to handle for a while -- leaves Caldera to concentrate on the low end where Linux and SCO 32-bit operating systems are jostling for the same mind share.

Allen said he doesn't see any problem with Caldera positioning and selling six different operating systems. But I just can't see it as a sustainable model. In my own opinion SCO blew its brains out by forever (it seems) selling multiple OS platforms that did many similar functions but were largely incompatible.

OpenServer, SCO's cash cow for the latter part of the 90s, can't do a single thing that UnixWare or Linux couldn't do better, faster, and/or less expensively. But VARs who have installed OpenServer have a lot of hours invested in systems and methods that have been on auto pilot for many years. Caldera will need to provide OpenServer tools on Linux that will allow for relatively painless migration.

UnixWare is another question mark since its role as the high-end Unix OS for Intel-based systems will vanish the day that Itanium and the IA-64 version of AIX L start shipping. UnixWare's Linux capabilities are superior to AIX's, but I wonder what UnixWare's appeal will be as AIX grabs for the high end, and 32-bit Linux encroaches on its high-end features.

And then there's IA-64 Linux, which could take a run at AIX's turf sooner than anyone thinks while also helping Itanium-based systems to be affordable as IA-64 migrates to commodity hardware.

I agree with Allen that the IA-64 will prove cost effective compared to RISC, especially when it becomes available on commodity hardware. Yeah, so Alpha was here first with 64-bit. Big deal. The Intel platform will have a greater variety of OS choices, and -- more importantly -- won't be coming from a platform vendor whose Linux commitment is lukewarm at best. When added to the Linux support given that other commodity CPU vendor, AMD, I think it won't take long to make the open-source world forget about Alpha. Compaq had a great head start and did nothing with or for the Linux community, so to heck with 'em.

What this all means is that I think Caldera's wealth of OS riches give it plenty to work with in supporting what I believe will be a very popular hardware platform. Indeed, it almost looks like too much to work with. While I don't know if I'd offer Caldera the same advice to 'run like hell' from AIX L as my anonymous SCO friend did, it's clear that AIX isn't the long-term necessity for Caldera that it was for SCO. Bringing together UnixWare's technology and Linux's freedom on IA-64 may be a better way to get there from here in the long run.

Would you invest in IA-64? Tell Evan in the TalkBack below or in the ZDNet Linux Forum. Or write to Evan directly at evan@starnix.com.

I imagine that, by now, some folks at Caldera have noticed the fact that their company is involved with no less than six operating systems.

First there are Caldera's present Linux-based products, eDesktop and eServer. Linux IA-64 is the partnership formed to ensure that Linux gets the most from the IA-64 (Intel's 64-bit CPU architecture). And OpenServer and UnixWare are the two long-running Unix variants the company obtained when it bought SCO.

And then there's AIX L.

"But wait," I hear you ask. "Isn't AIX the name of IBM's proprietary Unix variant?"

Well, yes it is. Thank you for asking. And herewith we find what could rightly be known as The Operating System Formerly Known as Monterey.

Just about two years ago, SCO, IBM and a bunch of other well meaning Unix companies decided that it would be better to combine efforts than to continue to produce similar but competing Unix platforms. Code named Monterey, the project was designed to enhance UnixWare's 32-bit capabilities while working to have a 64-bit Unix ready in time for the release of the Intel CPU, then called Merced.

The Linux factor
It's an understatement to say that much has changed since then, most of it having to do with this Linux thing. IBM has jumped on the Linux bandwagon with both feet, both hands and a bunch of other body parts. SCO went through a love-hate-love evolution with Linux until it was recently purchased by Caldera, a Linux specialist a fraction of SCO's size that now has the task of juggling these six different operating systems.

Most interesting has been Monterey's evolution away from a project that was supposed to bring together the best of all vendors. While Monterey was originally hyped as a merged product, UnixWare administrators looking to migrate to AIX L are in for a surprise, while existing AIX installations won't have much problem at all. Monterey has appeared to become simply the next release of IBM's Unix, with an IA-64 port, some Linux capabilities, enhanced multi-processing and multi-platform UDI driver technology.

SCO's Monterey product manager, Scott Allen, has no problem with the IBM-specific naming or the project's general evolution into a hierarchy of companies (with IBM firmly entrenched at the top) rather than a cooperative partnership.

"AIX has broad recognition in the enterprise," Allen said, "and it's very important for the combined SCO and Caldera to have a presence in that market." Allen said that AIX was more advanced than UnixWare in features that are necessary to this highest peak of the market, and it has more available applications.

Furthermore, Allen said that the AIX recognition will be helpful when approaching the first potential market for IA-64-based systems. "Our market research shows the early adopters will be RISC users who want to improve their price-performance." Scott said IBM, SCO and others were aiming squarely at Sun's share of the large systems market.

According to Miles Barel, program director, AIX marketing at IBM, it's all going according to plan. "We have done exactly what we said from the beginning," he said. According to Barel, the new AIX is incorporating two UnixWare features -- the /proc filesystem and the System V printing system.

One of my own questions about all this was how SCO, a company committed to hardware vendor neutrality, was going to sell and hype an OS whose name was so tied to one specific player. Would companies such as Compaq and Dell, who have been pretty closely tied to SCO, be enthusiastic about selling an OS made and named by a direct competitor? According to IBM's Barel, "We decided to retain the name AIX on the advice of our customers." Well, I'd expect that from existing IBM customers. But wouldn't existing UnixWare customers and hardware partners get scared off?

Allen didn't think this would be a problem, because AIX L will give these OEMs a way to compete against RISC vendors in a way they haven't been able to do before. "Are they really excited about the product being named AIX? Of course not," he said. "Will the name prevent them from promoting it? No. This is mainly about making money and the OEMs know they can do well with AIX."

Not everyone within SCO is as enthusiastic. One who (understandably) requested anonymity volunteered this:

"SCO worker bees have been screaming about these horrors for most of the two years this 'partnership' has been in place. We're mystified as to how this relationship got this far out of control. From what I can tell, we exchanged several person-decades for the privilege of being an AIX VAR."

Strong words, but ironically the commandeering of Monterey by IBM may suit Caldera's interests better than those of the pre-Caldera SCO. Having IBM maintain AIX L at the very highest end -- in the thin air of 32-CPU systems and huge installations, which Linux won't be ready to handle for a while -- leaves Caldera to concentrate on the low end where Linux and SCO 32-bit operating systems are jostling for the same mind share.

Allen said he doesn't see any problem with Caldera positioning and selling six different operating systems. But I just can't see it as a sustainable model. In my own opinion SCO blew its brains out by forever (it seems) selling multiple OS platforms that did many similar functions but were largely incompatible.

To each their strengths
OpenServer, SCO's cash cow for the latter part of the 90s, can't do a single thing that UnixWare or Linux couldn't do better, faster, and/or less expensively. But VARs who have installed OpenServer have a lot of hours invested in systems and methods that have been on auto pilot for many years. Caldera will need to provide OpenServer tools on Linux that will allow for relatively painless migration.

UnixWare is another question mark, since its role as the high-end Unix OS for Intel-based systems will vanish the day that Itanium and the IA-64 version of AIX L start shipping. UnixWare's Linux capabilities are superior to AIX's, but I wonder what UnixWare's appeal will be as AIX grabs for the high end, and 32-bit Linux encroaches on its high-end features.

And then there's IA-64 Linux, which could take a run at AIX's turf sooner than anyone thinks while also helping Itanium-based systems to be affordable as IA-64 migrates to commodity hardware.

I agree with Allen that the IA-64 will prove cost effective compared to RISC, especially when it becomes available on commodity hardware. Yeah, so Alpha was here first with 64-bit. Big deal. The Intel platform will have a greater variety of OS choices, and -- more importantly -- won't be coming from a platform vendor whose Linux commitment is lukewarm at best. When added to the Linux support given that other commodity CPU vendor, AMD, I think it won't take long to make the open source world forget about Alpha. Compaq had a great head start and did nothing with or for the Linux community, so to heck with 'em.

What this all means is that I think Caldera's wealth of OS riches give it plenty to work with in supporting what I believe will be a very popular hardware platform. Indeed, it almost looks like too much to work with. While I don't know if I'd offer Caldera the same advice to 'run like hell' from AIX L as my anonymous SCO friend did, it's clear that AIX isn't the long-term necessity for Caldera that it was for SCO. Bringing together UnixWare's technology and Linux's freedom on IA-64 may be a better way to get there from here in the long run.

Would you invest in IA-64? Tell Evan in the TalkBack below or in the ZDNet Linux Forum. Or write to Evan directly at evan@starnix.com.