I think you were right to identify Opteron as a major threat. x86 had taken at lot of market from the UNIX/RISC vendors with Pentium Pro particularly before that, but Opteron really opened a new front in taking the fight beyond the low-end. There were of course big boutique x86 systems from Sequent and Unisys and the like, but those were more like the big UNIX systems -- expensive low volume custom parts. Opteron was the first x86 with scalable, glueless SMP that shared most IP with low end parts. It wasn't until Nehalem 5 years later that even Intel had a comprehensive answer to it either.
I think Scott was right about BSD being too late though. Linux just had the momentum at that point that the BSDs would never again match. Particularly in commerical and "enterprise" supported products. Oracle DB for Linux was available in 97 for example. Sun would have been pushing shit uphill to convince Oracle to support a new port for a BSD they would sell on x86 systems.
OpenSolaris was really the last straw response to Linux I guess, but that was entirely a case of trying to imitate competitors. Not that it may not have gone somewhere if it was opened earlier, but for a time they actually threw open a lot of their internal development forums and things I remember reading through some of them and their performance and scalability and feature targets for OpenSolaris were literally "try to close our gaps with Linux". It was too little too late.
OpenSolaris around Opteron timeframe might have been a little more interesting, although 2003 saw the release of Linux 2.6, at which point it had really built a good head of steam. So unlikely to have made much difference in the long term IMO. The iron would have been far hotter in the 10 or so years before that.
> Sun would have been pushing shit uphill to convince Oracle to support a new port for a BSD they would sell on x86 systems.
That would have been far and away the easiest part of that mission. Oracle will keep an utterly obscure port of the core database product alive for a single customer if they're paying enough money, or at least that is the way it was 20 years ago, and their other products generally just depend on a working JVM. A BSD port in cooperation with Sun would have been extremely easy.
Making enough people in Sun care about desktop computing again would have been the hardest part.
Regarding the part of post I think you're responding to (Oracle will keep an utterly obscure port of the core database product alive for a single customer if they're paying enough money, or at least that is the way it was 20 years ago), it certainly does seem possible the attitudes changed at Oracle in 2009. In a short timeframe Oracle bought Sun Microsystems, a hardware competitor to Itanium, and Red Hat (whose Linux distribution Oracle clones) announced they would not support Itanium in their next enterprise Linux release.
I think you were right to identify Opteron as a major threat. x86 had taken at lot of market from the UNIX/RISC vendors with Pentium Pro particularly before that, but Opteron really opened a new front in taking the fight beyond the low-end. There were of course big boutique x86 systems from Sequent and Unisys and the like, but those were more like the big UNIX systems -- expensive low volume custom parts. Opteron was the first x86 with scalable, glueless SMP that shared most IP with low end parts. It wasn't until Nehalem 5 years later that even Intel had a comprehensive answer to it either.
I think Scott was right about BSD being too late though. Linux just had the momentum at that point that the BSDs would never again match. Particularly in commerical and "enterprise" supported products. Oracle DB for Linux was available in 97 for example. Sun would have been pushing shit uphill to convince Oracle to support a new port for a BSD they would sell on x86 systems.
OpenSolaris was really the last straw response to Linux I guess, but that was entirely a case of trying to imitate competitors. Not that it may not have gone somewhere if it was opened earlier, but for a time they actually threw open a lot of their internal development forums and things I remember reading through some of them and their performance and scalability and feature targets for OpenSolaris were literally "try to close our gaps with Linux". It was too little too late.
OpenSolaris around Opteron timeframe might have been a little more interesting, although 2003 saw the release of Linux 2.6, at which point it had really built a good head of steam. So unlikely to have made much difference in the long term IMO. The iron would have been far hotter in the 10 or so years before that.