Solaris Conversion Rate: Sun Sheds Some Light

Published: November 15, 2007

by Timothy Prickett Morgan

Last week, when Sun Microsystems went through its financial report for its first quarter of fiscal 2008 ended September 30, Jonathan Schwartz, Sun's president and chief executive officer, said in a conference call with Wall Street analysts that the company has over 2 million processor sockets globally with paid-for support for the Solaris 10 operating system. This is the first time that Sun has given any hard numbers that allow us to reckon a conversion rate from Solaris 10 freebie shipments to Solaris 10 paying customers.

That conversion is, after all, what Schwartz's plan is for Sun as he transitions the company from sales of closed products to distribution of free and open source products that have support fees. Sun wants to have a vast installed base of customers playing around with Solaris Unix--particularly on its competitors' iron--and is betting that that a percentage of these customers convert to for-fee updates and technical support when they roll some of their applications into production environments on Solaris. This is exactly the same bet that commercial Linux customers make, of course, and while this hasn't made any of them billions of dollars in sales, as Unix operating systems used to, Red Hat can hardly be said to be an unprofitable and unsuccessful company in terms of its finances and influence in IT, and Novell would have been bankrupt--spiritually if not financially--had it not bought SUSE back four years ago and started becoming a player in commercialized Linux.

Schwartz said in the call with Wall Street last week that as the first quarter closed, the company has shipped nearly 11 million cumulative Solaris 10 licenses worldwide. So a conversion rate of more than 2 million processor sockets could be a large conversion rate. If you assume the average server has two sockets, that is 1 million servers, and that is a 9 percent conversion rate from freebie to paid license. Sun originally offered basic support for $120 per socket for updates and patches, $240 per socket for 9x5 business support, and $360 per socket for 24x7 support. In January 2007, when Solaris 10 Update 11/06 started shipping, Sun changed its Solaris support prices and packaging. Now, the basic support contract provides 30 days of telephone support and one year of online support for $240 a year on a two-socket server. A standard 12x5 support contract--that's 12 hours a day, five business days a week--costs $720 on a two-socket machine. A premium 24x7 contract costs $1,080 on a two-socket box, and it offers mission-critical support for Solaris as well as any application that is running on Solaris. At the low end, developers are being offered a $49 annual email-based support contract, and enterprise customers are being offered a custom annual contract for their sites called the Solaris Everywhere plan. Under this plan, Sun stops counting boxes, and it only makes sense for customers who have dozens or hundreds of machines. Assuming the Solaris 10 average customer paid somewhere in the middle of each scenario, Sun is generating hundreds of millions of dollars in Solaris 10 support fees--probably twice the rate per machine since the prices are roughly twice as high under the new scheme compared to the old scheme.

Of course, that is theoretical Solaris support money. In practice, it is a lot harder to say exactly what Sun has achieved with Solaris 10 in terms of conversion and how much it is charging. The original Solaris support prices cited above, for instance, were only for Solaris 10 support on X86 and X64 servers, not on Sparc boxes. And Sun bundles Solaris with its own servers, so it can allocate as much or as little of the system cost to Solaris licenses and support as it sees fit.

Here's the problem in trying to suss this out. Every one of Sun's own Sparc-based and X64-based servers has Solaris bundled on it, and so do Sparc servers from Fujitsu-Siemens, so these machines are part of that cumulative Solaris 10 install number--and have been since Solaris 10 was launched in January 2005. Those install figures also include Solaris 2.7, 8, and 9 customers who plunked down Solaris 10 on their existing iron, and about one-third of the shipments (or around 3.5 million machines) for Solaris 10 were for Sparc boxes. If you count new Sparc and X64 server shipments since January 2005, that's about 600,000 machines from Sun, and maybe another several tens of thousands from Fujitsu-Siemens. Assuming the average machine has two processor sockets, that would yield around 1.2 million processor sockets sold by Sun and Fujitsu installed bases since Solaris 10 was launched. Not all of these sockets were running Solaris 10, of course. But for the sake of argument, let's say 1 million of them are. That would mean Sun and Fujitsu-Siemens could account for half of the Solaris 10 sockets under support contracts just by the sheer fact that they bundle it on their boxes. And the existing installed base of Solaris customers could easily have consumed support services for another 1 million processor sockets and barely dent the Sparc-Solaris installed base, which is probably around 1.5 million machines and maybe as much as 3 million to 5 million sockets worldwide, even after massive consolidation. (It is hard to guess these numbers, really.)

In other words, for all we know, Solaris has brought a lot of new or returning customers to the Solaris 10 fold in terms of using the operating system, but the same old core Solaris shops might be paying for the services Sun peddles for Solaris. We just can't tell from the limited information available. What seems reasonable is that if Sun had a 50 percent Solaris 10 conversion rate and was adding hundreds of thousands of new, paying customers to the Solaris support fold, it would have long since said so. But again, who can say for sure? All IT vendors are cagey about sales statistics.

What Sun has suggested in the past is that it has a software business that is worth in excess of $1 billion, and it is a good guess that only a small portion of that comes from its licensing of Java technology, its Enterprise Java System middleware stack, its myriad system and development tools, and its support contracts for the OpenOffice desktop suite. Most of this money, however Sun carves it up, has to be coming from Solaris. And the real measures Wall Street wants to see are how many paying Solaris customers are out there, and are they driving growth in Sun's software business? Sun has promised to break out sales figures for its software businesses, but before it does that, the company has to figure out how it will allocate hardware, software, and services for a bundled product. As is the case with all IT vendors who sell all three, the differences between hardware, software, and services is being smeared. Sun is trying to give clarity and transparency exactly when it is most difficult to do so.

RELATED STORIES

Sun Wrings Profits from a Flat Fiscal First Quarter

Sun Finally Gets Solaris 10 11/06 Update Out the Door

Solaris 10 Breaks Through 6 Million Shipment Barrier

Solaris 10 Hits 3 Million Registered Licenses

Can Solaris 10 Shipments Continue Upwards?

Sun Talks Up Solaris for X86, Misses Big Opportunity

                     Post this story to del.icio.us
               Post this story to Digg
    Post this story to Slashdot