4 ms·

Define sucking. Sun is the only midlevel server company that innovates beyond buying mother boards and CPUs and assembling them. They make their own CPUs, thei

by andr 18y ago

Define sucking. Sun is the only midlevel server company that innovates beyond buying mother boards and CPUs and assembling them. They make their own CPUs, their own OS, ZFS, serious storage solutions, etc. They foot the bill for MySQL and Java.And so on and so on. Unfortunately, bad management can sink a good technology company.

wheels 18y ago
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.
- jaaron 18y ago
  "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).
  - wheels 18y ago
    At that rate, MySQL should pay itself off for Sun by 2025. Any bets on if Sun will be around? ;-)
- crabapple 18y ago
  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)
  - pjjw 18y ago
    there are quite a number of them, especially in academia. the support contracts are ludicrously expensive in some departments i work with.
- rbanffy 18y ago
  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. I hope they do. An x86-only world is dull.
  - lallysingh 18y ago
    Thoughts on the x86 Solaris workstations? <$900. IMHO, the Niagra chips are nice, but are fairly specialized vs desktop chips. They're pretty well set up for lots of threads, but I donno how they'd do on the desktop. They're still sparcs, after all.
    - rbanffy 18y ago
      That's the problem: You shouldn't care about how well a SPARC desktop runs Firefox. You should use such a box to develop software that run on a 32 or 64-thread machine. We must be prepared to deliver performance on many-core machines because that's what the desktops and servers will be in the near future. And that's why Sun should investigate cheap desktops that could be used to run and develop - they have a uniquely viable platform for that and they have it today. Any software built for those boxes will run better on their servers. As a personal favorite, they should work their asses off to make sure "fashionable" languages like Python and Ruby do threading decently on those chips. If I have to select a web server and I have a US$ 20K server with a 4-thread x86 and a US$ 20K 32 thread SPARC, I will probably go Sun's way.
- donw 18y ago
  "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.
  - crabapple 18y ago
    tell me one concrete example where a hard limitation of linux kept you from meeting your goals, yet opensolaris got you across the line
  - wheels 18y ago
    I meant all of Sun's competitors, you know, those guys that are eating their lunch. The distinction here is between the quality of their product and their understanding of their market.
whatusername 18y ago
AIX and Power6 doesn't count for IBM? Surely big blue can get a little credit there.. Agreed though - I'm not really sure what HP do with Servers - more expensive than Dell, and all off-the-rack components...

by andr 18y ago

wheels 18y ago

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.

jaaron 18y ago
"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).
- wheels 18y ago
  At that rate, MySQL should pay itself off for Sun by 2025. Any bets on if Sun will be around? ;-)
crabapple 18y ago
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)
- pjjw 18y ago
  there are quite a number of them, especially in academia. the support contracts are ludicrously expensive in some departments i work with.
rbanffy 18y ago
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. I hope they do. An x86-only world is dull.
- lallysingh 18y ago
  Thoughts on the x86 Solaris workstations? <$900. IMHO, the Niagra chips are nice, but are fairly specialized vs desktop chips. They're pretty well set up for lots of threads, but I donno how they'd do on the desktop. They're still sparcs, after all.
  - rbanffy 18y ago
    That's the problem: You shouldn't care about how well a SPARC desktop runs Firefox. You should use such a box to develop software that run on a 32 or 64-thread machine. We must be prepared to deliver performance on many-core machines because that's what the desktops and servers will be in the near future. And that's why Sun should investigate cheap desktops that could be used to run and develop - they have a uniquely viable platform for that and they have it today. Any software built for those boxes will run better on their servers. As a personal favorite, they should work their asses off to make sure "fashionable" languages like Python and Ruby do threading decently on those chips. If I have to select a web server and I have a US$ 20K server with a 4-thread x86 and a US$ 20K 32 thread SPARC, I will probably go Sun's way.
donw 18y ago
"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.
- crabapple 18y ago
  tell me one concrete example where a hard limitation of linux kept you from meeting your goals, yet opensolaris got you across the line
- wheels 18y ago
  I meant all of Sun's competitors, you know, those guys that are eating their lunch. The distinction here is between the quality of their product and their understanding of their market.

whatusername 18y ago

AIX and Power6 doesn't count for IBM? Surely big blue can get a little credit there.. Agreed though - I'm not really sure what HP do with Servers - more expensive than Dell, and all off-the-rack components...