Thursday, December 01, 2005

Sun rising?

Sun announced yesterday that it's making its entire suite of server-side software available under an open source license. John Loiacono provides some background. Jonathan Schwartz posted a teaser the day before. In fact, if you read Loiacono's post to the end, you'll see he promises more big news on December 6 -- Sun's beginning to preannounce its announcement dates regularly.

If you've been watching Sun for a while, this is no surprise. OpenSolaris clearly telegraphed the intent. Sun's been talking about releasing Java under an open source license for a long time. Schwartz's blog post links to several other earlier statements that this was going to happen.

So what does it mean?

If you're a customer, it means you can suddenly get Solaris and the applications that run on it under the same terms, and at the same price, as you get Linux and its applications. Sun's using a different open source license than the one used by GNU/Linux, but that difference matters less to IT managers than to guys like me. In practice, if you run an IT shop, you can now download, build and install the server software you need to run your business. You can choose between Fedora from Red Hat (or one of the other freely-available Linux distros) and OpenSolaris. Linux will have to compete more aggressively on features, reliability and scalability with Solaris. Customers can be happy. Competition and choice are good, and the price is certainly right.

Of course, Sun expects to sell subscriptions to IT shops running Solaris, just as Red Hat and Novell do for their Linux customers. Sun clearly hopes that the freely-available software will convince developers to build high-value systems relying on Solaris and Sun hardware, which will drive adoption of the platform more broadly.

If you're a Sun competitor, the picture is muddier.

Among proprietary vendors, Microsoft probably doesn't care much. It already had exactly this headache with Linux, and Microsoft will say that Solaris is competing for market share with Linux, and not with Windows Server. (Sun will certainly disagree). IBM has an established open source strategy and practice based on Linux already, and plenty of muscle in the marketplace to acquire and retain customers. Hewlett-Packard is likely in the worst shape among the proprietary Unix vendors. It has no clearly-articulated open source strategy, depends on hardware sales for a substantial fraction of its revenues, and will have a harder time than ever differentiating itself from IBM, Sun and the commercial Linux distributors. (HP's current struggles are distressing. Let's hope that a great company can recover its earlier success. The garage in which the company began is undergoing restoration now for eventual opening as a museum).

Open source vendors like Red Hat and Novell will be publicly dismissive, but privately concerned, about the development. Without question, Solaris will win some deployments that would earlier have gone to Linux. The key question is how many, and I can't predict the answer to that question. If the losses are mostly to roll-your-own shops that don't buy subscriptions, well, it's a shame, but not a crisis. If Solaris begins to pry away subscription contracts from Linux vendors, though, life will get interesting in Raleigh-Durham and Provo.

If you're a Sun shareholder, you hope this works. Sun's stock closed down twelve cents yesterday after the announcement, at $3.77 per share. The company is a long way into a big transition from a big-iron vendor with a lot of proprietary high-end software, to a nimble player offering commodity systems and freely-available infrastructure on a subscription basis. To make this work, Sun needs to end its reliance on large up-front equipment purchase fees, and to compete on price with vendors of Intel and AMD-based systems. Sun needs also to switch from a model that recognizes large up-front license fee payments for software, and instead move to a subscription-based revenue stream where bookings are recognized ratably over the course of the year.

Even if Sun can reestablish leadership, operate profitably and drive new growth, the cash flow, top line and income in the new business look very different from those of ten years ago. The market isn't yet assigning any significant value to the new operating model. If it begins to drive revenue to the extent that Wall Street notices, the questions will become: What is the appropriate earnings multiple for the new Sun? How does it compare to the multiple for the old Sun? Should a smart investor buy at $3.77? The jury is still out.

For Sun itself, this looks to me like exactly the right move. The decision to put Jonathan Schwartz into the key operational role was a good one. It was time to let a software guy run the company. The widespread adoption of open source development and distribution models meant that the big vendors like Sun, IBM and HP had to figure out a way to participate, before they were rendered irrelevant. Sun's making excellent progress in transforming itself. I hope that it works.

The move isn't without substantial risk. You'll see, if you read Schwartz regularly, that despite my branding him as a software guy, he runs a hardware company. When he talks about value to IT managers, he talks about delivering computer systems that use chips from AMD and Intel, but that consume less power, generate less heat and take less space than competing boxes. Jonathan claims that the main problem in the data center is the electric bill, not the software bill.

Sun's position is interesting. In the 1960s, IBM leased computer hardware to its customers, and installed and managed the software on it for free. Beginning in the 1970s, and continuing until just about today, the model was different: Buy a box, buy some software, and hire some people to keep it running. Sun is returning to a model more like the 1960s. The computers are much nearer zero dollars than they were in the 1960s, but you pay a vendor an ongoing fee to keep your combined hardware and software running, and to make sure you have the latest software all the time. Sun's not alone in this, of course. The whole move to hosted applications and subscription-based IT services do the same thing. Sun is the clearest example of a hardware vendor doing this, though.

Worth saying: I'm not a Sun shareholder. Sun's a long-term and excellent customer of my company. I was an early fan (I was at Berkeley working for Bill Joy when he left to found Sun with Andy Bechtolsheim and Scott McNealy). I have been skeptical in recent years that Sun could remain a long-term leader in a rapidly-changing technology market. Long-term leadership is hard. I'm awfully interested to see what happens next.


Blogger Mike Olson said...

Jonathan Schwartz posted an interesting item on this announcement at

11:29 AM  
Blogger Rex Wang said...

I agree with you, Mike. Sun has been telegraphing this move for quite some time -- making software free and open source to drive adoption. That enables them to sell subscriptions for coherent, bundled stacks and support, and of course, drives sales of hardware. Sun software has always been a loss leader for hardware, so this seems like a very natural progression to me. Free software with paid support, all driving the real business: low price / high performance hardware.

2:57 PM  
Blogger Rex Wang said...

Here is Scott McNeally talking about razors and blades in Business Week:

"Well, the software is the razor. The razor blades are the servers, the storage, the memory, the service contracts, the archiving services, the tape cartridges, the integration, the consulting services."

3:55 PM  

Post a Comment

<< Home