Sun president talks databases, Sparc and HP
- 23 February, 2005 08:56
Hoping to lure new customers with open source software, Sun Microsystems has now added a database to its planned line-up of open-source products, joining its Solaris operating system and other products. But what's in it for customers, and who's going to care?
Jonathan Schwartz, Sun's president and chief operating officer, was in France last week for the 3GSM World Congress in Cannes. IDG News Service and Le Monde Informatique, an IDG France magazine, caught up with him at Sun's offices in Paris to talk about its open source plans, Sparc chips and other products. He also had some advice for Carly Fiorina's successor at Hewlett-Packard:
What's the main benefit for your customers of making Solaris open source, and which types of customers will it appeal to?
We're obviously serving our existing customers with the existing Solaris. Open source Solaris is a means of reaching new Solaris customers, and customers who have decided to run on HP, Dell, IBM or any of 400 other equipment makers' hardware. Customers have articulated a preference for running open source software -- customers who want to interact with the source code and maybe modify it to experiment with performance, or just out of curiosity. It really opens a world of new customers.
Do existing customers benefit, too?
Of course. There is no downside to being an open-source operating system; transparency is a good thing.
What types of new customers do you expect to attract?
Given the pricing model attached to it -- which is free -- it's a wonderful way of appealing to a broad new set of markets. The barriers to adopting Solaris have been reduced to zero. As to which markets it will open, I think everything from the Brazilian government, which has expressed a preference for open source software, to academic environments that want a good teaching platform.
Does Sun have a concrete plan to offer an open source database, or was Scott McNealy just being provocative when he suggested that recently?
To be a complete application platform you have to have some form of persistent storage. You can achieve that through a file system, a directory engine, a messaging store, the persistence engine in our application server -- those are all forms of databases. What we haven't done is address the SQL access database, which has been served well in the open source community by MySQL and PostgreSQL. We're committed to filling the hole -- all of the hole, not just the file system. We have to address the requirements of the SQL database, so I think we're quite serious about it.
Would you use the same model as you did with Linux on the Java Desktop System, i.e. take an existing open source product, tweak it for your needs and put a Sun label on it?
That's to be determined. Customers have said, 'We'd like an alternative to the existing choices we have.' And they are consistently asking Sun to go work on that issue.
So it's a matter of when and not if?
If Solaris for x86-type processors turns out to be popular, doesn't that mean a decline in your Sparc business? How do you balance the two?
About one fifth of the downloads of Solaris 10 within the first 10 days were onto Sparc. About four fifths of the downloads were onto x86. Of those downloaded onto x86, the vast majority were to non-Sun hardware. There are two ways to look at that. One is that the downloads to non-Sun hardware were lost hardware opportunities. My way of looking at it is that downloads to non-Sun hardware are now leads for our sales force to begin penetrating non-Sun hardware accounts. To me it's all about growing the market.
What proportion of those half million downloads were by existing Sun customers?
I would bet that the majority were not Sun customers; the downloads over the Web tend to go to developers and smaller customers first.
In the longer term, won't existing Sparc customers turn to an x86 platform like Solaris-Opteron for cost savings?
Well, the good news is that, given we're the Number 1 Opteron supplier in the world, we have a good shot of winning their business. Customers always have a choice and it's a good thing.
Every product at Sun has to win on its own merits. Do I believe we can win a 64-way Sparc system against a comparable IBM power system or Itanium system? Unquestionably. Do customers have the choice to run that on an x86 Opteron platform? Absolutely. Sparc is perfect for some jobs and for others it is not. Similarly for Opteron. If you wanted to run a 64-way Opteron system today you'd be out of luck because there's not one made.
So how will you compensate for the loss of revenue from Sparc when more customers start running Solaris on Opteron?
You're making an assumption that is incorrect, which is that customers prior to the release of Solaris x86 did not already have the choice of using x86 instead of Sparc. All we've done is enhance our ability to earn new revenue.
But in the past you weren't promoting Solaris on x86.
But that's a means of capturing that market, not a means of leaking revenue. My last blog entry is all about a meeting with an ISV. They had elected to move to x86 because they wanted to move to a low-cost platform with one- and two-way servers. They left Solaris for Red Hat. What they told me was they are moving back to Solaris, and by the way, would we like to bid on the hardware? The leakage, to the extent there was cannibalization of our existing business, has been going on for years.
You talk often about Red Hat as if its your big competitor, but many ISVs now also support SuSE Linux, thanks to Novell's involvement. Hasn't the market changed a lot?
The number-one competitor for Solaris is not Red Hat, it is Windows. We see that market now opening up. Open source Solaris gives us access to huge new market opportunities. No one has to lose for Solaris to win. What we all expect to see is a widescale build-out of Internet infrastructure, where Solaris' value proposition around scale and security and real innovation will give us a leg up. For us to win no one needs to lose, we just have to grow faster than the rest of them.
How do you react to Gartner's prediction that by the end of the decade, Itanium and Power will each have about 15 percent of revenue from the chip market and Sparc will have only half that much?
On a unit volume basis we're far and away the number-one 64-bit microprocessor platform. We've consecutively grown and taken share from our competitors. How do I feel about my friends at Gartner who like to forecast the future? Well, they missed the browser, they misforecast Solaris shipping half a million units in the first 10 days of its release, and they predicted Microsoft would succeed wildly on handsets. I enjoy their perspectives but they are not necessarily ones we share.
What's the value-add of Sun's Opteron and Xeon boxes, is there any differentiation from IBM and HP?
On a one-way box I doubt anyone would be very differentiated on the hardware itself. The single biggest advantage we have on one-way up thru n-way is that we ship an operating system, and to the extent that you subscribe to that operating system we may be prepared to take the price of the box under its cost. At which point I'd ask the same question to Dell and HP: How do they expect to compete? If their box is US$1,000 and Sun's is $700, then for customers most interested in price we're $300 more interesting.
Speaking of HP, if you were made its chairman tomorrow, what would you do?
I'd break the company up. The printing division generates the vast majority of the value in the company but there's absolutely no synergy between a printer and an Itanium server, so why would they depress the value of the printer company by attaching a boat anchor of a systems business?
So what would you do with the enterprise and PC businesses?
Find a more capable vendor to take them forward. I just don't think it has stand-alone value
How about if they bought some middleware, say BEA Systems?
A failing systems vendor that's acquiring other companies is still a failing systems vendor. BEA does not make an Itanium server any more attractive than it was before BEA.
Jean-Luc Rognon is an editor with Le Monde Informatique, an IDG News Service affiliate.