As executive vice president of Sun's software group, Jonathan Schwartz heads the company's new unified software business and is leading the charge to promote the Linux open-source technology. He sat down with Steve Gillmor and Tom Yager to discuss Sun's recently announced Linux desktop strategy and to explain how it's in the industry's best interests to develop an alternative to arch-rival Microsoft.
Can you bring us up to speed on where Sun is, particularly with your Linux desktop announcements?
Schwartz: There are a series of dynamics in the industry that are yielding some interesting disruptive opportunities. To the extent that we have always been a little bit of the disruptive innovator, these announcements are in some ways a very natural evolution of something that we're already doing. The architectures that we've established for the past six years write to Java on the server, write to a browser on the front end, write to J2ME-enabled clients, and use Java Card. All of your applications are available through a browser. For the most part, you don't write to Windows anymore. We're at a natural fruition and the culmination of this architecture where we can now "port" all those applications to a new desktop. Why don't we go ahead and do that and decouple the insecure, closed, and proprietary system that causes everybody a lot of heartache, not the least reasons for which are licensing and security and system administration and complexity? This is a great opportunity to ask: is there enough momentum, maturity, technology, and quality available in an open-source desktop to get us to where we need to be? With Mozilla, Evolution, StarOffice, Java2, and Linux, we can deliver a secure desktop that allows folks to not have to necessarily upgrade off NT 4.0, which is going to be abandoned at the end of the year. We give people a way of getting out of the cycle of pain associated with software assurance. We give them a very low-cost alternative to what exists.
What we announced at the SunNetwork conference is a server running portal, messaging, directory, identity, and Solaris -- or Solaris X86 or Linux. And it's free. That's part of what we announced. For the first hundred users, all the software we're talking about is free. On the client side there’s Linux, Mozilla, Gnome, Evolution for e-mail, and a Java Card reader in a secure PC that's OEMed from one of six Asian suppliers. By the first quarter of next year, we expect to have customer-ready systems, which at Sun means you get a crate and in the crate are all the systems. You take them out, plug them in, turn them on, and everything works.
So you agree that the browser is not dead?
No way. If you look at what the .Net framework is, Microsoft goes around and says, "The submit button is a very inefficient mechanism for interacting with the user. You should really use the .Net framework." There's only one bug in that system: As soon as you use the .Net framework, all of a sudden your service is now available only on Windows. You can't run it on a Linux desktop unless we go figure out a way to reverse-engineer the .Net framework.
What's the relationship between the low-cost Linux LX50 server bundle you announced several weeks ago and this new desktop strategy?
The relationship is Sun ONE -- it is the idea that you should write applications to Java and to Web service standards so that they can be re-purposed for any device attached to the network. If you want to do that on an LX50 running Linux, great. If you want to serve 1.5 million consumers, I have a hard time seeing how you're going to do that on Linux.
For developers, isn't there a disconnect between a server model based on Java and a client model based on Linux? Are you talking about them converging?
No. Nor am I talking about writing to C. People don't do that anymore. Linux is an operating system, it's not a developer platform. Linux is a tactic. Java is the strategy. The developer platform that we're encouraging is for line-of-business applications, content-based applications, distributed applications. Java is the architecture. It runs on the highest-end carrier-grade servers and it runs on the military-grade, most secure smartcard microprocessor platform on the planet.
But how's that going to play on the Linux desktop?
We will integrate Java Card into the J2SE platform. The one bug in the system right now is that for the most part, the Java platform and the Web content worlds have diverged. It's incumbent upon us in a Web services way to cause them to converge.
Isn't this reminiscent of a speech Bill Gates gave years ago at a Microsoft Professional Developers Conference about bridging the gulf between Win32 and the browser?
We run J2ME on all the smartcard platforms. Java runs on all the set-top boxes. Ninety-five per cent of all microprocessor smartcards are Java Card. It runs [on] Amex, the United States Department of Defence and the whole country of Taiwan. So I've got a lot of momentum right now, unlike what Bill Gates was doing, which was to say that we have to give up something to get there. What I do is I add something in. I get more security, I get more interactivity, I get more authentication. What's interesting to me is the desktop PC is the only unauthenticated network access point left. I can walk up to any PC anywhere and go send a virus around. A big part of the initiative for us is the integration of Java Card. Our initiative is going to be rooted in a secure PC, because that security mechanism is going to become a critical differentiator. The strategy is: let's authenticate the last portion of authenticated systems, as well as add into our line-up of clients the one that's been a little difficult to get, because the host has been a little recalcitrant.
Is there an analogy on the client side to the Control Station controller box on the server side? Absolutely. There will be two versions of these clients. One is called Sun Ray, which is a solid-state client. If you unplug it from the wall, it's a piece of plastic, it's useless. So the FBI no longer has a problem of worrying about people taking their computers home with them. You can't take this one home with you. The user management there allows us to just dramatically reduce the number of system administrators associated with managing desktops. Managing a Unix network of desktops is much, much easier than managing a Windows network of desktops, which is why in general the comparisons are 1:500 for Unix and 1:100 for Windows.
Comcast has to manage 1.5 million consumers. For me to manage 100 users is very, very simple. I already do it today on Sun Rays, and I already do it for Messaging and Directory and Portal, and all the other things that'll run on a desktop. So we will use the distributed user management system that's already available in Sun Ray for these Linux desktops. It's not very complex. If you look at how a Linux desktop is configured, there's a little configuration file that tells you what's available, when, for whom, what documents are open, where the windows are. That's all just balled up and we're going to hopefully get that on a little smartcard, so as you walk from Linux PC to Linux PC, you'll be able to take [that information] with you. We can also do things in a Linux desktop environment that are really, really difficult to do in a Windows world. For example, you can't use Visual Basic. You can't add any file to your desktop. You cannot access your local drive. There's a bunch of things that we can do that are relatively difficult to do.
What about heterogeneous environments of desktop clients?
All of this stuff needs to run compatibly with Windows, which is why StarOffice is there, which is why Evolution is there, and why Mozilla is there. How do we cooperate on the networking side? We do what we do today, which is provide PC NFS and support for interoperability between Unix and Windows environments. Consider the security hassles alone -- there have been 60 Microsoft patches this year. That's a lot of money. My belief is we can eliminate almost all of that.
What if HP wants to do that?
You have no idea how opportune that question is. We have discussions with all of the major PC OEMs going on right now because they all want to talk to us. The thing that's a little weird is we have nothing to lose. This requires no real incremental investment and we're going to polish this up and make this much easier to deploy and manage. We don't have to worry about eroding our PC margins. If this succeeds, our margins will rise because more people will want our portal, directory, messaging, calendaring, and identity services. If this succeeds, we will ship more storage to contain all the mail messages, pictures, and what have you that are in that application.
Are you talking about freeing people from subscription-based software plans, where you never finish paying for what you buy?
No. The only real issue I have with software assurance is the terms and conditions under which the users were expected to migrate. You sign up or we're going to screw you to the wall. And secondly, the fact that users had no negotiating power because there was no choice. They couldn't really run to a Mac. I wish they could have. I love OS10. So at Sun we are likely to have a subscription model for those customers that want it that way. We will also have a full purchase price. I was with the CIO of a satellite broadcast company who told me his average cost was between 600 and 700 bucks per user, per year, on software alone. And he's running a call centre. I said: "I guarantee you, no matter what, we can come in under that." Now you may not have exactly the same [apps] as you have currently, because StarOffice isn't exactly Microsoft Excel, and Gnome isn't exactly Win XP, and Evolution isn't exactly Outlook, and Mozilla isn't exactly Internet Explorer. But I'm not sure your call centre employees need those.
Is your goal simply to reduce Microsoft's margin to the point where they have to back off?
I'll tell you the things we're not going to do. I don't want this on Amazon. I don't want to have a consumer product. I don't want to piss off Yahoo because we didn't pick their home page as the home page on our browser. Or an infrastructure company. We ship systems. Would I love to see Yahoo and Sprint and Comcast and AOL Time-Warner label these products out into their value chains? Absolutely. Am I going to go fight over what icon they can put on the desktop or tell them that if they take off Mozilla and put on Netscape, we're not going to give it to them? Absolutely not.