Sun set on server business?

Sun set on server business?

Summary: Companies that have run Sun hardware now know they can continue to get competitive gear even as Moore's Second Law continues to bite.

TOPICS: Oracle, Hardware, IBM, Servers

Thomas Watson Sr. with THINK sign 200 pixelsIn all the hullaballoo over Sun's agreement to support Solaris 10 on IBM hardware I have yet to read one obvious fact.

This is part of Sun's exit strategy from the server business.

(The picture of IBM's Thomas J. Watson Sr. is from the Computer History Museum.)

I mean no criticism in saying this. It has been obvious for years that companies like Sun simply can't compete in the chip market any more.

The reason is Moore's Second Law.

Everyone trumpets Moore's Law, the idea that chip complexity increases exponentially, doubling every 18 months or so. But just as complexity increases exponentially so do development costs.

Sun has recognized this fact through its open source hardware efforts. By sharing design specifications, costs are shared. The life span of designs are expanded as custom versions are developed. It's a true win-win.

This IBM deal is as well. Companies that have run Sun hardware now know they can continue to get competitive gear even as Moore's Second Law continues to bite.

It's a smooth glide path, a real contrast to what others have faced when Moore's Second fell on their hardware suppliers. (I happen to know people who are still running Unisys gear.)

Jonathan Schwartz has done a great favor for his own customers, and may increase software revenues down the road. But all his spin is really lipstick on a pig.

In many ways, Sun is becoming Red Hat.

Topics: Oracle, Hardware, IBM, Servers

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • Great post Dana - hadn't considered this angle

    Made me think more broadly about the deal. I don't think Sun is exiting the server business. They're just expanding the market reach of Solaris.

    • Aw shucks

      Another intelligent poster noted that "servers are where the money is," which is why Sun isn't admitting it's getting out of servers.

      But if everyone else makes your hardware, and you can't get the volumes on your own chips to be competitive on price, you're letting yourself be driven out of the business, whether or not you care to admit it.
  • Sun said from the start

    that they wanted Solaris x86 to be OEMed by IBM, HP and Dell and OS support is a large market.

    And they just released the UltraSPARC T2, and they will release the UltraSPARC RK sometime next year. Also, they sell SPARC and x86 chips from Fujitsu and Intel respectively. Assuming they don't make chips, I don't think they'll ever stop selling servers - coz that's where all the money is!!
    • Think Apple

      Apple had to move among three chip-makers -- Motorola, IBM, and then Intel -- because its design requirements meant it couldn't get the volumes it needed.

      The same is going to happen with Sun, at least in terms of basic server chips. By getting its software onto other chip sets, and onto other platforms, it maintains its relevance.

      But it becomes less a server company and more a software company. An open source software company.
  • RE: Sun

    Sun's hardware was never up to snuff. I could install an Amdahl or IBM mainframe and have the customer up and running in 8 hours, a Sun/Fujitsu SPARC server took most of a week to get there. This is still a good move for Sun as far as Solaris is concerned, on the hardware side, they can concentrate on x86 hardware hoping they can compete with IBM, a very tough ask.
  • Sun chips are OEM'd from TI and Fujitsu

    Sun assembles their servers with chips from many OEMs.
    This is also true for HP and Dell.

    Sun does not manufacture its own SPARC chips.
    They get them from Texas Instraments and Fujitsu.

    Sun does not manufacture its own x86 chips.
    They get them from AMD and Intel.
  • Not so obvious.

    I think that the reason that no one else had written this "obvious" fact is that it isn't obvious at all. I don't even think it is true. First, the cost of design of chips isn't exponential like chip complexity. If it were, then no new chips could be made because it would cost in the trillions of dollars.
    Second, just because Sun is expanding their software doesn't even suggest that it is contracting the hardware. Sun has already shown that OS innovations still matter, and now with the latests chips they are also showing that innovations in hardware matter as well. Why would they plan to exit the server market just after announcing the highest performing server chip on the market? And third, your argument is generic, applicable to all chip designers. By that reasoning, everybody should exit the server market. No, Sun is clearly expanding in the server market and the IBM deal is to help expand server sales, not replace them.
  • Complete nonsense

    Sun spent $2 billion developing the T1 processor, and probably at least that developing the T2. Why would they do that if they were about to exit the hardware business? You've taken 2 plus 2 and made 5!
  • Sun *do* design SPARC processors

    You're half right. Sun design their SPARC processors, but since they don't own any fabs, they get TI to manufacture them.
  • what a load of nonsense!

    Dear Dana,

    you could not be more wrong -- try a new set of glasses for those tea leaves -- or new tea leaves?
  • The Silicon Valley Biz Model du Jour

    1) Hire engineers to make quality hardware
    2) Open source your software and make it available on your quality hardware
    3) Make your competitor's open source software available on your quality hardware
    4) Make both of these scenarios available as a virtualized software as service
    5) Provide both remote and onsite professional staff to help extend the software for your clients
    6) Tap (pay directly) the open source development community for enterprise support for your clients
    7) Explain to institutional investors why company resources are targetted at the engineers, professional staff and enterprise support folks first, C-Suite second. (Talent, not corps, matter).
    8) Hope institutional investors aren't so clueless as to miss the paradigm shift. If you happen to have your retirment funds with them, you might take it upon yourself to help educate them.