As a former systems engineer at Sun Microsystems, Daniel Berg had a front-row seat during the development of Java. Today he is the chief technology officer of Cyrus Intersoft, which plans to release a set of Java system services this month, called Speiros, that are designed to let any client be linked with any major server platform. In a conversation with Michael Vizard, Berg talked about what it will really take to fulfil the original vision of Java.
IDG: What's your mission?
Berg: Cyrus Intersoft was put together to deliver what we're calling pervasive network computing, meaning that anytime, anywhere, anyone could go to any type of device and access its resources. If you look at what the Web did, the Web made content pervasive. The problem we saw was that although the Web is very good at deploying content, it's not necessarily the best for deploying applications. So we put in place a platform that would allow applications to be delivered in the same way that content is delivered today. Right now you're looking at the Internet through one window, and that's the browser. What we want to do is bring the resources of the Internet down to the user at a level that they're most comfortable with, which is to click on a file and launch it.
How does Speiros work?
The server side is really a collection of servers, depending on which pieces of the system you choose to implement. The server pieces were done in 100 per cent Java so that they can run across different platforms. Everything we do is Java; there's no native piece of anything anywhere, so we are truly platform-agnostic. If you look at Java the language or the Java platform, it gives you everything you need to have an application written in. But what it doesn't take into account is the systems aspect of it. What about user accounts and user security, and how am I going to do remote printing and remote file systems?
So what differentiates you from Sun or any other platform vendor?
If you look at Sun with its release of i-Planet, it's kind of the same thing in the sense of anywhere, anytime computing. But they're again restricted to the browser, and quite frankly, the browser is a really lousy application engine. A lot of these companies have tried to do this thing, but what they tend to do is base technology upon existing technology. If you base it on existing technology, then as you move forward you're probably going to have that as a crutch and it's not going to work everywhere. Sun's i-Planet is an example of that. It's a pretty good solution, but it only can be served from Sun systems. Obviously that's what they want, but it ties the technology down and doesn't open it up.
Anything else that's unique about your platform?
Metering, or software rental, if you will. There is really nothing like it today, but if you look at the direction things are going in, why am I going down to Egghead Software and buying [Microsoft] Word and only using 15 per cent of its capabilities and only using it 10 per cent of the day? Why can't I get that capability that I want, for a tenth of a cent every hour or something like that? That capability is built in as well, the metering aspect, so you can meter really anything, whether it's single-use, minutes of use, bandwidth use, file-system use. That is transparent to the application.
From your perspective, what's been holding up the whole notion of network computing?
If you look at how some of these companies have tried to implement it, they've tried to take the hardware approach. One of the big problems with Java that we've cracked is that right now, each Java application needs its own virtual machine in order to run. What we do is divide the name space in the Java virtual machine in such a way that we can run multiple applications in the same VM. A Sun JavaStation could execute one VM at a time. So if you had to swap between applications, you had to actually almost reboot in order to get it to bring up another application. Now you can bring up this environment as the VM, and the applets and applications resources can just run in that same environment.
Why did you have to leave Sun to accomplish this?
I'm a believer in the technology, and I'm a developer at heart. I think every developer believes in writing code that can be used from anywhere and not having to be locked in anywhere. That's really what Java tried to do, but I think it failed in a number of areas. Vendors are going to tie it back to technology they have today, and I don't think anybody has really taken this approach before. Sun is still a hardware company. They're going to tie things into hardware, and i-Planet is a great example of that. They have anytime, anywhere computing, as long as the server is Sun.
What impact have the Java wars had on Java's momentum?
Well, I came from Sun, so I'm a little bit biased. But I think it's really growing. What hasn't happened is the promise of write once, run anywhere. It's sort of now write once, test everywhere.
So what's going to make people stand up and take notice of you guys?
We're really targeting the developers at JavaOne [a conference in San Francisco this month]. We're giving away 100,000 shares of stock in the company to get them to come by the booth and take a look at the technology. You have to stop by the booth, pick up a certificate, and then go to the Web site. We're going to have contests for different applications that are deployed.
Cyrus Intersoft is at http://www.cyrusintersoft.com