Interview: Marimba seeks fellow players

Arthur van Hoff, CTO at the world's hottest Java start-up, left Sun Microsystems along with his co-founders earlier this year believing they could help grow the market for Java. Now, Marimba is part of a movement that some say will revolutionise the way Internet and intranets are used on based on software tools that automatically deliver applications and content to users .

Arthur van Hoff, CTO at the world's hottest Java start-up, left Sun Microsystems along with his co-founders earlier this year believing they could help grow the market for Java. Now, Marimba is part of a movement that some say will revolutionise the way Internet and intranets are used on based on software tools that automatically deliver applications and content to users .

Marimba's entry into the emerging "push" software technology market is Castanet, a system that includes a "tuner" that users download and store HTTP-compatible server plug-ins that work with tuners to deliver Java-based software applications or Web content, called "channels." Van Hoff, on a visit to Japan, spoke to Rob Guth of the IDG News Service.

IDGNS: What kind of corporate culture are you trying to foster at Marimba?

ARTHUR VAN HOFF: We definitely have a very strong vision that we want to have a dynamic and fast moving company. We don't want to get stuck on a long development cycle. We don't want to get behind. The Internet is moving very fast and therefore you need to create a company that can move very fast and also change directions very fast. I think our biggest example in that world would probably be Netscape. They have very, very fast release cycles although I don't think we would want to grow as fast as Netscape. Growth is very difficult - you have to be very careful that you don't grow too fast because it can really change the culture of the company. That's why we like to work with big companies, because they create lots of leverage.

IDGNS: What hardware platform do you have in mind for Castanet? It seems your concept needs a hard drive so do you look at the JavaStation or NC or the hard drive-based NetPC being promoted by Microsoft and Intel as a target for your technology?

VAN HOFF: We believe that all Internet computers and network computers need some kind of local storage. A hard disk is one option. Another option is static RAM or anything like that. The reason is that bandwidth isn't free. Also as these devices become more mobile the connectivity becomes more fragile. You either have to use a modem, which is slow and you need a local cache, or you are not always connected. So if you are walking around on the street what do you do if you don't have local storage? Local storage is an absolute requirement for successful network computers in the long term and I think that Castanet fits very well into that model.

Putting a hard disk into a JavaStation has two drawbacks. One is that it is the only component with moving parts which brings down the lifetime of a machine ... and secondly the user has to administer the hard disk. With Castanet we solve that problem: the hard disk is essentially administered remotely by a system administrator or by a developer who provides the channel.

IDGNS: At Comdex, Netscape announced that it will use Castanet in its Constellation component. What other kinds of applications are you targeting with the technology?

VAN HOFF: There are a whole range of different types of applications you can build. You can think of the PointCast-like applications to distribute news or weather but you can also think of more MIS-focused applications such as travel expenses, account management or sales information database access. There is a real wide range.

The reason why there is a lot of excitement about Castanet is that it answers a lot of problems. It really enables developers to start developing real applications in Java and have a way of distributing them. Today if you try to build an MIS application and you are restricted to HTML and Java and JavaScript, it is very complicated and very unreliable. If you want to do a mission-critical application then you're much better off looking at something like Castanet because in the end you will find that something based on HTTP is actually fairly fragile and also it creates a lot of overhead on your server.

IDGNS: Primarily the basic model for your partnerships will be some vendor embedding your technology in their product?

VAN HOFF: There are a number of possible ways we can work with people. There is desktop integration. That's an obvious way where we create more clients. There are alliances for distributing our server technology as part of other server technologies. We can talk to Internet service providers for building Castanet-ready Internets and intranets.

We are looking at network computers. We have lots of opportunities in the network computer world. We are looking at handheld devices and consumer devices. A company that produces audio/visual equipment that wants to install software upgrades for it ... we have the problem that we have too much opportunity. There are too many ways that we can go, so our challenge today is to remain focused.

IDGNS: There are a lot of companies that are promoting the "push" concept of Internet content and software distribution. Marimba's technology also fits into this category. Over the next year what will users see, a flurry of companies like yours all trying to position themselves as the single provider of push technology?

VAN HOFF: That is an interesting question. I would hope that a standard will develop that people can use. We position ourselves not as a competitor to these companies, more as a solution provider. For example, PointCast. A lot of people mistake them as a competitor. We are a technology company and they are a content company. We would like to provide technology to them that they can use in their business to create their own channels. I do think that the user experience on the Internet will become more and more TV-like. There's much more aggregation of knowledge or of information for the user. Also personalization, I think, is becoming more and more important.

IDGNS: What about a companies like Intermind Corp. or Backweb Technologies? Do you see them as content companies as well or are they competitors?

VAN HOFF: Something very similar to Backweb could be built on top of Castanet in a very short amount of time. We would like to see them as a customer, although I don't think they're very interesting in being our customer right now. They are obviously trying to promote their technology as the right answer. But their focus is different. Our focus is very much on software distribution. We think that the term software or the term channel is a superset of a lot of solutions out there and we solve the generic problem first and companies like Backweb and Pointcast - all of these companies - can build on top of those solutions more focused solutions like ‘how do I distribute a background pattern for my Windows 95 thing?' ... or ‘how do I distribute a news flash?’

IDGNS: Just a clarification - you say "software distribution" - do you see the technology you provide as focusing primarily on distribution of software and not content?

VAN HOFF: When we talk about software and when we talk about a channel, that is code and data. A product like PointCast is some of amount of code and some amount of data and we ship that as an entire package to you. And we make sure it's always up-to-date so we update the data and we update the code. So the lines are very blurry. There's a very thin line between software and content.

We're very good at delivering software or data that changes once an hour or a couple of times in an hour, but it's not real-time. As soon as you start talking about real-time, then you want to use different types of protocols because our protocol is really optimized at doing very efficient, low overhead updates that are at most a couple of times in an hour. Whereas, if you wanted to have latest stock quote up to the minute, you would use proprietary stock ticker protocol or you use a database access protocol. But that doesn't mean that you couldn't deliver the database client software with Castanet and have it make a live connection to the database.

IDGNS: It has reached the point that perhaps for an upstart company to make it these days it needs to partner with Microsoft. Can you give some details on any negotiation you may be having with Microsoft?

VAN HOFF: We can't comment ... You are sort of alluding to a problem with Java in general: If you develop a library as a third party then you almost have to get a license or deal with Netscape or Microsoft in order to distribute your software. There are lots of companies that are doing that today. They are becoming partners with one or the other. It really makes the market very black and white. You're either pro Microsoft or against Microsoft. I think that it is very much a shame.

IDGNS: What about the possibility of joining or becoming part of a larger entity such as Netscape or Microsoft?

VAN HOFF: When we started our company the four founders decided very early on that we wanted to sell products, not companies. We've had a lots of offers and we turned them all down. We think that if we hang in there and build good products that we'll become worth much more and be much more productive than if we join a bigger entity.

IDGNS: Are you considering a public stock offering?

VAN HOFF: That's very premature. We want to have revenue and establish our company very firmly before we do that. We do not see going public as a source of revenue ... A lot of people mistake going public as a source of revenue. They think that's how you make money in start ups. I think you make money by selling products.

Join the newsletter!

Error: Please check your email address.
Show Comments