Well, but if nobody else sees value in producing their own CPUs for mid-level servers, maybe there's a reason?
If everybody else is using a free OS for mid-level servers, maybe there's a reason?
Java, OpenOffice and MySQL are neat. Know anybody who's paid money for them?
Sun's decisions in the technology world have been so bad in the last decade that you could have become rich just by watching what they do and investing in the opposite.
"Java, OpenOffice and MySQL are neat. Know anybody who's paid money for them?"
Yes. Java still makes money for Sun for embedded and mobile devices. The licensing costs for a mobile JVM are reasons why you see the rather unique approaches taken by Android.
Back in 2006, MySQL had $60 million in revenue, so someone's paying for it. Not sure about OpenOffice (or Star Office, that is).
Java, OpenOffice and MySQL are deffensive strategies. If corporations didn't see a viable option (and Linux was no such thing 7 years ago), even more of them would have gone the Windows route and Sun would have seen its high-margin Solaris/SPARC business collapse long ago.
Workstations with x86s pretty much eroded their low-end Solaris/SPARC market because even a low-end SPARC cost twice as much as a pretty decent x86 running Windows NT. That was the largest mistake - to allow NT to take a foothold in the low-end workstation market. This caused lots of sophisticated software being developed for Windows instead of Unix.
It's probably too late to develop a minimalistic low cost, sexy, sleek box running OpenSolaris out of a T1 or T2 chip, but I would consider such an all-or-nothing strategy to put lots of very cheap, conceptually advanced computers in the hands of students and researchers so to foster development of software that runs better on Sun architectures.
It's a good kind of vendor lock-in: one achieved by excellence. Sun is uniquely positioned to do that.
IMHO, the Niagra chips are nice, but are fairly specialized vs desktop chips. They're pretty well set up for lots of threads, but I donno how they'd do on the desktop. They're still sparcs, after all.
That's the problem: You shouldn't care about how well a SPARC desktop runs Firefox. You should use such a box to develop software that run on a 32 or 64-thread machine.
We must be prepared to deliver performance on many-core machines because that's what the desktops and servers will be in the near future.
And that's why Sun should investigate cheap desktops that could be used to run and develop - they have a uniquely viable platform for that and they have it today. Any software built for those boxes will run better on their servers.
As a personal favorite, they should work their asses off to make sure "fashionable" languages like Python and Ruby do threading decently on those chips. If I have to select a web server and I have a US$ 20K server with a 4-thread x86 and a US$ 20K 32 thread SPARC, I will probably go Sun's way.
"Everybody" is not using Linux for their mid-level servers; Solaris, especially OpenSolaris, is still a very viable option, and offers a number of features (most notably ZFS and Zones) that Linux isn't really close to, yet. Not to mention, Solaris has, in my experience, a much better I/O scheduler.
Linux is a fine operating system, but it isn't the end-all-be-all.
yup. sun is a bespoke tech firm in an era of relentless commoditization.
my guess is that they only cling to life on support contracts from people who are still solaris/sparc shops (which i imagine they are still quite a few of)
If everybody else is using a free OS for mid-level servers, maybe there's a reason?
Java, OpenOffice and MySQL are neat. Know anybody who's paid money for them?
Sun's decisions in the technology world have been so bad in the last decade that you could have become rich just by watching what they do and investing in the opposite.