% fortune -ae paul murphy

AIX vs Solaris (5)

The "threatscape" concept says in part that things which affect key suppliers affect your costs - i.e. that the possibility that Exxon might be motivated to buy out and shut down Sun has cost implications for the data center manager choosing whether to bet the business on Solaris/SPARC or AIX/Power.

Sun, of course, is a lot bigger than Zilog was and there are a lot more constraints on directors trading dirty deeds, so this isn't a real threat - but other factors may be.

On the SPARC/Solaris side all the threats affect the company, not its product choices. Within Sun the commitment to Solaris/SPARC is pretty absolute - and open sourcing just about everything has reduced the benefit a big competitor might get if someone launched a financial attack on the company. As a result the biggest current risk to Sun, and thus to SPARC/Solaris, is that something goes seriously wrong during the current transition to CMT/SMP computing - but a bet that looked pretty scary for Sun three years ago now looks like a done deal with the T2 exceeding expectations, Rock pretty much on track, and Solaris 10 setting new standards for both technology and reliability.

Because most third parties writing for Solaris cover their bets by ensuring portability across both Solaris for x86 and Linux, vendor lock in is as close to a non issue for the majority of Solaris/SPARC applications as you can get.

You do incur lock in, however, when you use unique Sun products like Sun Rays, start to depend on unique Solaris functionality like SMF, or build your data center to T2 SWaP levels. It comes down to this: if you want to use the new stuff that makes Solaris/SPARC a world beater, there are no second choices - meaning that a real problem at Sun could lead to technology abandonment and significant change for you.

With IBM most of these concerns are reversed. Thus corporate threats to IBM are all internal, but the commitment to AIX seems weak and I believe that problems with Power6 performance are mainly a consequence of software lag and an expected part of a planned across the board transition to Cell computing.

Application lock in is more of a threat with AIX than Solaris, largely because there are fewer vendors and, as a group, they tend to be more conservative. On the other hand IBM's increasing emphasis on Linux is pushing many of them in that direction and thus opening competitive opportunities - and reducing vendor lock in risks for users.

AIX is well behind Solaris and even Linux - and that has a risk reducing benefit because there's really nothing significant you can do with AIX that you can't switch to something else provided you're willing to abandon AIX specific hardware and minor AIX features in products sets like those from Tivoli.

More dangerously, however, the SCO case casts a long and unpleasant shadow over AIX. That case, fundamentally, is about the claim that IBM failed to honor its contractual obligation to maintain the confidentiality of information contained in, or derived from, licensed AT&T code - and was therefore initiated by an SCO letter lifting the licenses under which IBM offers AIX and related products to its customers.

That issue has not been settled - and evidence that IBM allowed one experienced AIX programmer to help with the mainframe Linux project at the Bvblingen skunk works would probably settle it for SCO.

Notice two things: groklaw will tell you that the case was about copying, but that's wrong -the whole threat to Linux thing was a legal maneuvre designed to establish a fair market value for SCO's claim against IBM. Groklaw will also tell you that the issue is over and gone, but that's wrong too: whoever ends up having the right to enforce the contracts, will also end up with a duty to do so - meaning that a Novell director, if that company turns out to hold the rights and obligations involved, cannot meet his responsibilities to shareholders other than IBM by leaving a potential billion or two in easy money on the table.

Notice further that this is a threat to AIX, not Linux - meaning that a data center manager adopting Linux need not care about this case at all, but one considering AIX should consult the company lawyers to get a good understanding of how real the risks are and what the cost consequences might be.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.