Sun targets Red Hat, not Microsoft
- By Joab Jackson
- Feb 16, 2005
Jonathan Schwartz, Sun president and CEO
Is there any major IT company under more competitive pressure right now than Sun Microsystems Inc.? Sun's specialty has always been network computing products and services, so it should be living large in this emerging age of net-centric services. Yet, the Palo Alto, Calif., company has been getting heat from all sides. Linux is being touted as a low-cost alternative to Sun's Solaris operating system. Commodity x86-based processors are being marketed by competitors as an alternative to Sun's own high-end SPARC processor-based servers. All this flack has program managers in the federal government'one of Sun's largest customers'worried about its long-term viability.
Sun president and chief operating officer Jonathan Schwartz doesn't seemed worried, though. In charge of Sun's day-to-day business affairs, Schwartz has a compelling vision for the company's future. Schwartz holds degrees in economics and mathematics from Wesleyan University.
Holding forth in a media roundtable at the Washington Post's Washington headquarters, Schwartz was interviewed by Washington Technology senior editor Nick Wakeman, Washington Post technology policy reporter Jonathan Krim, Post technology editor Larry Liebert and GCN associate writer Joab Jackson. GCN: What is Sun's Linux strategy?
SCHWARTZ: You mean what is our Red Hat strategy? My belief will be that there will be more systems running Linux next year than ever before. They will run wristwatches, set-top boxes, embedded systems, network equipment. It will be everywhere. And there will be no one Linux, because the diversity of Linux in the client world will correlate to the diversity of clients. And that diversity is a good thing. It creates lots of competition.
Now in the data center, diversity is not a good word. Most businesses can't afford the diversity. Many of our Wall Street customers write a bunch of Linux applications. In fact some of them write their own Linux distributions. Then they go to an [independent software vendor] and say 'I need you to write your application to my operating system.' The ISV will say 'No, we only support Red Hat stuff.' So what is happening to Linux in the data center is that the market has tipped to one company, which now has raised prices and basically presented a very proprietary operating system, called Red Hat Enterprise Linux. [To read about Red Hat's push in the federal market and its latest release, Red Hat Enterprise Linux 4, see our Jan. 24 issue, Page 10, or visit www.gcn.com
, GCN.com/369.]GCN: How will Sun address this situation?
SCHWARTZ:Look at what Hewlett-Packard is doing. HP has said it is going to deliver volume x86 systems and run Linux on them. HP has largely end-of-lifed HP-UX, because it doesn't run on x86.
What we have done with Solaris is quite the opposite. In 1993, when the marketplace told us to abandon Solaris and take up Windows, we elected to redouble our efforts on Solaris, which allowed us to build a customer base. Rather than throwing that away to pick up a Red Hat product, we moved our OS to other platforms.
So now, if you look at volume x86 platforms, there are only three OSes to pick from'Windows, Red Hat and Solaris. The good news is Sun owns one. The bad news is that Dell doesn't own one, HP doesn't own one and IBM doesn't own one. They may have to acquire one.
The most interesting thing about Linux, aside from the social movement aspect, is the fact that it is the first Unix to run on x86. So we put Solaris on the x86. Now we have well over a million licenses of Solaris running on Intel and on the AMD Opteron.
So, when we talk to a customer and they say they have a bunch of Linux applications, we say we can run them natively, with no modifications, on Solaris 10.GCN: With the porting of Solaris to x86, are you worried about making money in a commodity market?
SCHWARTZ: Let's talk about the hallmarks of commodities. The good thing about commodities is that they are demanded by everybody on the planet. So you never have a demand problem with commodities. Oil, gas, electricity, water, power are all commodities. The other hallmark of a commodity is that there is a transparent price: a nickel a kilowatt hour, a buck a gallon, 45 bucks a barrel.
I think there has been a lot of great theater made from some companies that say IT is commoditizing, yet this [holds up a cell phone] is no commodity'being about $2,000 worth of phone I got for $100 when I signed up for a three-year contract. What is commoditized isn't the device. What is commoditized is bandwidth. Bandwidth has been the only thing that been commoditized. Now if you really believe that a four-way Opteron server running a robust Linux system with a Java enterprise middleware stack is a commodity, then your definition of commodity is really quite different from mine.GCN: What will be the user benefits from Sun's 2004 partnership announcement with Microsoft?
SCHWARTZ: First and foremost, interactivity around our directory infrastructure. You can now take a large Active Directory implementation and interoperate with Sun's Java Enterprise systems operating on Solaris. That gives us a unique interoperability that other market players do not have.
Longer-term, identity really is the most fundamental Web service. You have to start from there. The question is what can we layer on top of identity management'anything from digital rights management, database performance, application infrastructure.GCN: So five years from now, how much of IT will be hardware? How much will be software? How much will be services?
SCHWARTZ: What do you think the price of oil will be? My favorite analogy to that is this: Five years from now, how much of Verizon's revenue will be handsets, versus capacity or service? It doesn't matter. The revenue derivative of the product itself is unrelated to the value the customer buys. How much does Verizon make on its handsets? They lose about 15 percent of their revenue on handsets. That's interesting. So according to the [logic of the question], Verizon should shut down their handset business. But you don't want to shut that down because that's where you get all those subscribers, who want new handsets. So I think in our business you can't look at a specific product and say how much revenue did it bring? It's not a useful question.GCN: How much technology spending will we likely see this year? Will we see growth?
SCHWARTZ: That really depends on which audience and which set of issues you are talking about. There were a billion wireless devices sold last year, 10 times the number of PCs. There will be more technology bought that people will carry around in their hands than ever before. So in that sense, the demand for services and service operators will grow.
I think one of the dangers, especially in the media, is that all of us in this room are of a particular demographic. We are no longer the target demographic of the Internet by any stretch of the imagination. How many of you have ever bought a ring tone? [One hand goes up.] Ring tones are about a $3 billion industry.
So this definition of the network represented by a PC is really nothing related to what the network is on. If you ride JetBlue, you get to watch satellite TV in the seatback in front of you. You get a sense that the network really is where people are. I think today we are seeing the evolution of a bunch of different industry transitions [to make] the world replete with mobile network devices.
Verizon is doing an awful lot of upgrading now. Is Bob's flower shop? No, but I think there was a lot of IT that was built for Bob's flower shop that never should have been built. There is a huge number of people that IBM has employed by custom building everything from customer relationships management systems to order-entry systems. Now you're better off just going to SAP, or better yet to [hosted CRM service] Salesforce.com.
We've been spending a lot of time with financial service and gas companies. Every one of them are building grids. Everybody is building a grid, and they are almost all identical. So we said, why don't we build a multithousand-CPU grid and we'll sell it for a dollar a CPU-hour? And there has been an overwhelmingly positive response from CIOs.
There's a few folks alienated by that approach, among them all the people who are building grids. They don't like the concept that they could be replaced by a buck-an-hour charge.