As the United States enters the post-bubble, late-recession, prewar (as I write this) economy, one of IT’s biggest money machines, Sun Microsystems, is lobbying to sell you something. Anything.
They’ve got your big monolithic server, your little $US1000 Sparc server, your blade server, and your rack-mountable PC. They despise Intel, but you can stuff Sun-branded x86 blades into Sun’s new blade chassis. Sun’s big on Solaris, but they’re just as big on Linux, and they’ll grudgingly embrace the existing Windows servers in your datacenter. J2EE? Sun’s got that.
A slide from the February 10 Network Computing 03 press conference lists the cost of software for a Sun 12-processor server as $0, which must be the permanent price tag for Solaris 9 and Sun ONE (Open Network Environment).
And, finally, Sun’s new N1 Provisioning Server makes network configuration so easy that one guy running a Web browser can do the work of a whole IS team. Sun is simultaneously eliminating complexity and building up its services offerings to outsource the complexity it’s eliminating.
Competitors will portray Sun’s moves as desperate. I don’t think Sun would deny that. Was there ever a better time for a technology company to pull out all the stops? If IBM, Hewlett-Packard or Microsoft chuckle at Sun’s mad R&D scramble, customers should ask Sun’s competitors why they’re not doing the same thing.
Sun understands that for at least the next year, companies are going to be making basic decisions about how to spend limited IT budgets. There are some market forces Sun can count on — server and manpower consolidation come to mind — so no waffling is necessary on offerings such as blades and N1, which target those areas. Beyond that, it’s anybody’s guess what IT will deem important. The lesson Sun learned from IBM is to bet huge on your favourite horses and bet a little on each of the rest. True, some horses are darker than others. The odds on Solaris x86 making it to the first turn must be 200/1. The potential payoff makes the bet worth putting down, but Sun must be willing — and I believe it is — to let customers kill off this and other Sun technologies that don’t meet their needs.
Sun CEO, Scott McNealy, has committed to spending $US500 million per quarter on R&D. That’s pulling meaningless — read, “conveniently round” — numbers out of thin air, like the $US6 billion Sun claimed to spend in 2002 giving StarOffice to people who can’t afford computers. Still, I admire a public company that ups its R&D burn rate in the face of declining share prices. Stockholders would reward Sun for decimating its workforce and selling off some of its prized assets. But Sun knows its customers are selfishly focused on what Sun can do for them, even if it means Sun might go out of business trying to deliver it. If it doesn’t try, Sun certainly will not be able to compete against entrenched commodity suppliers such as Dell and Intel nor against go-for-broke competitors such as IBM and Microsoft.
The company that always worked so hard to educate — read, “pressure and lock in” — its customers is now willing to listen. To Sun’s credit, it understands that in this context — trying to reach the top tier of IT buyers in a down economy — listening is a costly and interactive exercise.
Most customers can’t sketch out exactly what they want from Sun. A customer will know the right solution only when he or she sees it, so the massive spending is Sun’s way of restocking its showroom.
This isn’t a sudden move. The surprise is that Sun has followed through on the high-risk plans it has telegraphed over the past several months. Last year, Jonathan Schwartz remarked that Sun’s strained relationships with JBoss and Apache — open-source Java projects — needed fixing. Within weeks, Sun mapped out a new open-source-friendly JCP (Java Community Process) model. It delighted standards and open-source hawks but ruined Sun’s chances for huge margins on commercial Java and J2EE licenses. In the long run, it’s a shrewd business decision that mirrors Apple’s approach: chop away at Microsoft’s enterprise software share by offering similar functionality for free.
Sun CTO, Greg Papadopoulos, said in January that dynamically provisioned networks of affordable, low-power computers would be the systems of the next technology epoch. McNealy followed two weeks later with breakthrough pricing on blades and storage hardware. Sun is pushing hard to keep either HP or IBM from dominating the blade/grid/utility space.
Not everything came off according to plan at the Network Computing launch. McNealy’s crisp presentation was blunted by embarrassingly primitive demonstrations of Sun’s visualisation and grid solutions. In another demonstration, Sun showed off its SSL proxy blade, a hardware encryption accelerator.
Hardware encryption is built into network adapters and consumer 802.11 access points. That’s not special enough to have its own blade; it should be built into the Gigabit Ethernet switch Sun put in its blade chassis.
To get some laughs from the crowd, a Sun exec poked fun at IBM’s gargantuan dual-processor x86 blade by waving around Sun’s slender alternative. The trouble is, a dual-processor Sun x86 blade was not among the products announced on February 10.
There will be other quarterly gatherings and lots of chances for Sun to tighten up its technologies and its message. Sun closed out the first-quarter event by saying, “Wait until Q2.” We will see unfinished and potentially short-lived technologies at each of these events, so a seasoned IT leader’s natural reaction will be to hang back.
But enterprise IT shouldn’t pass up this chance to tell Sun what matters and see how it reacts. McNealy talked with us about closing the feedback loop, reducing the time it takes to bring customer requirements to life as products. That, in one sentence, is the method to this apparent madness and disarray. Maybe it is crazy to hand $US5 billion to a gang of geek geniuses and tell them to find out what makes customers happy.
I can’t think of a smarter or saner way to spend that money.