ADT Q&A: David Litwack, SilverStream Software

Many a tools start-up describes its plans to be "the Powersoft of the Internet." Such firms may be in for an unexpected challenge - Powersoft founder David Litwack has said good-bye to PowerBuilder and joined SilverStream Software Inc., maker of a purely Java-based Web Application Platform that provides a communications software suite and agent-based "push" technology, as well as the requisite database access, all in an Internet-friendly environment.

Application Development Trends Managing Editor Jack Vaughan met recently in Burlington, Mass. with Litwack, just as the SilverStream president and CEO was coming off a SilverStream beta user meeting.

Yours is a new company that has done some aggressive 'betas' and your just coming off a mini-conference with your users. Generally, what did you learn from their feedback?

Well, we had close to 300 people which is pretty extraordinary for a product that hasn't quite started shipping yet, and the demographics were interesting. We had major, major corporations - Fortune 500-, even Fortune 200-type corporations. We had a broad set of very high quality resellers who are primarily a sort of industry-guru-consulting type of company. Then the third thing, we had a number of international people from as far as Australia, Japan, Scandinavia, Eastern Europe and so forth. So, it's a pretty broad group. So I think it says several things about the market - one is that the Web market for people who want to build serious business applications based on a Web type technology is extremely hot. If I were to compare early days of client /server to now I'd note there's no missionary selling going on - people understand what needs to be done. In many cases they've been trying to use existing technologies such as CGI 'apps,' they've been trying to build serious database oriented business apps. It's pretty much your typical early market where they 'roll heir own', buy a lot of glue, use several products, and be your own integrator.

So people are knowledgeable, people have immediate needs, people are not just at the pilot projects; they have several immediate projects that are waiting for the right technology. A second comment is we are not seeing the traditional geographical [breakdowns]. In the mainframe days, we used to talk about one to two years until a technology market was ready in Europe after North America. Today it's an instant simultaneous worldwide market. In some countries, the Internet is actually thriving faster because it is a partial replacement for a weak phone system. They're more likely to use E-mail than the phone because it's more reliable.

The third thing is that from a channel point of view, we're living in a market that - well in the mainframe days there was very little channel, you had to get your support from your hardware vendor. In client/server days, you had the birth of a channel and it grew up to be quite sophisticated. Today you have a market beginning that is starting off day one with a mature channel. It's all these small, medium sized and large consulting companies all of them are looking to transition to the Internet. So, you have the characteristics in the Web market of something that historically would have been occurring two or three years into a market rather than pretty much the beginning of the market. And I want to stress, when I say the beginning of the market - some people say the Internet started two to three years ago - that was phase one of the Internet. That's static Web pages, that's publishing of information online. When you talk about serious, interactive, database-oriented business applications we're at he very early stages.

You had 300 people at your beta user meeting because of the background of the heads of the company...

I think it's that and I think this is [the] right market. People have been looking for this type of product for a year now. So, I think that the company has a strong reputation, but if we had decided to get into artificial intelligence, I don't think we would have had as many people. One of the channel partners made the comment to me that a channel has to converge around two things - it has to converge around a technology and the channel has identified Java as the technology they want to converge around, But you can't converge just around a technology, a channel also has to converge around an institution of some sort. You need a company to converge around that will have channel programming, that will train partners, that and that will support partners and all these kind of things.

So what is an institution here?

I'm talking about a vendor. So, if I asked 'who is the Java vendor?' - well, it's not really Sun - Sun maintains the standard. A company that is a standard bearer has to have a couple of things. It has to have obviously a product that is appropriate for the market. And we have to distinguish things here - the third-party channel is not going to succeed and be profitable by doing applets and Web sites! They're going to succeed by doing business applications. They're going to leverage their skills in database [technology] and business applications design. The channel we're talking about is a corporate channel. These are the consulting companies that are in Aetna and in Merc and in Citibank - they know they have to move to the Web, they've anointed Java as the technology, and now they need a vendor to rally around. [So our draw] is a combination of the reputation of the people but also that the market was already looking for a product and company like this.

The Internet tools market has yet to prove fruitful. Do you worry when you seen a venture capitalist like Ken Fox describe it as a slaughterhouse space?

Well, first of all, let's talk about what it takes to do business applications. HTML by itself doesn't do it. Now when you start out to create dynamic pages, the first phase is CGI and anybody can do CGI. There's a first generation of Web tool vendors - what they did was they automated CGI. In other words, they gave you a user interface, wizard-based way of defining the rules to get at the database and generate the HTML pages. The problem with that is it's too limiting. It gets you dynamic pages but it doesn't get you a really good user interface on a client, it doesn't get you interactivity, it doesn't get you a true three-tier architecture. So, I think we represent the next generation of Web tools, and there are several things that characterize that next generation.

The single basic most visible one is client- and server-side Java in an integrated three-tier architecture. By that very statement you change the nature of things. First of all, you're not a tools vendor you're an applications server provider. You have to be, because to do client- and server- side Java, you're not going to build these huge Java applications that run on the client, and you're not going to do database access from the client. In a proper architecture, database access occurs from the server. In fact, the thought of the unknown browser user doing direct access to your database is pretty scary. So, you have a three-tier architecture where the client and server are extremely intimate with each other in getting the job done. And you have Java running on a client that presents a rich user experience, high degree of interactivity and a user interface that is at least competitive with client/server. So that's one of the first things to do with the next generation.

The second thing is in this world, rich content and data come together. So in client/ server you have the data world as typified by things such as PowerBuilder and Visual Basic, and you have rich content as typified by Notes. In the Web, people expect to see those combined. They don't expect to see those as discreet roles. And if you look at some of the things Notes brought to the table like universal access and replication and things like that, some of that you get automatically with the Web. We talk about rich user experience, well, the Web has brought very high expectations for rich user experience when it comes to unstructured information - documents, images, multimedia, and so forth. But it's brought very low expectations in terms of structured data. We all no the famous HTML form with the 'Submit' button, it's not the sexiest thing we've ever seen. So what people want to take is the very strong structured data capability they had with client/server and the very rich content experience with the Web and combine those. And if you think of the way human beings behave - in our minds we don't process information by separating out. Human beings process information as a combination of data. And, each paradigm shift that we have is a step in the evolution of human machine interface. And one of the things the Web is doing is bringing together those two things - structured data and rich content.

And the third thing - client/server was client centric, the Web is mid-tier-centric. A lot of your capabilities are going to reside in a middle-tier server. One of the consequences of that is you have a piece of software that knows everything that is going on. And so it can see mail coming and going it can see modifications to the data tables, it can be the integrated mail systems. And this naturally leads to 'push technology' which we do with our [software] agents.

But that doesn't address the question on the nature of the tools market.

Well, see, I think the mistake is it's not a tools market, it's a platform market, and the tool is a feature of the platform.

Well I guess that's why there's a lot of emphasis at SilverStream on the applications server. Could the application server become more important that the database server?

Well that's a pretty interesting question. I guess it would be hard to say 'yes' because the strong application server is going to be good news for database vendors because it's going to drive the demand for more databases. For example, SilverStream saves all our metadata, all of our rich content, as well as traditional data, in the database. So, I think it enhances the opportunity for database vendors and it will be good news for hardware server vendors. Can it be as big an opportunity as database servers? Let me phrase it this way: I think it's closer to the database server model than it is to the development tool model. But it's a new animal. The industry is creating a category called application servers.

For better or for worse, the Internet has been closely associated with consumer mass markets. These may allow some developers to demand cheap, or free, tools. You plan to make money on the server side. Did you have any choice?

First of all, the target market is business, not consumers.

I understand that.

Anybody, whether it's a software company or anybody, I don't care what the business is, the impact of the Web on your business, is staggering. We'll get an article, and get 1,000 leads in the next 48 hours over the Web from all over the world. We have partners that get e-mailed our press release automatically because they subscribe to a news service, and they have pressed the key word Java, they get on our Web site and download our white paper and e-mail us requesting a reseller license. Not only did that have no human contact, it didn't produce any paper! So, yes, we're a software company, but I think the Web is having an impact on the fundamental business model of every form of business. That's what driving us. This is not people frantic for the right technology; this is people frantic to make use of the Internet and change the fundamentals of their business. And we're seeing things like that whether it's an or a Federal Express, that's the tip of the iceberg.

Well, we've seen surveys where 'cost' comes up as the number one issue for Internet tools. That's something new. You're used to getting a demo off the Web site and the next step is not paying for it.

Well, the term Internet tool by itself is a broad term. For example, any HTML authoring environment is an Internet tool. And there's an expectation. Even the Java development tools are a totally different category of product because they don't have an environment - it's all roll-you-own stuff. But one of the things that characterizes business applications is to build a real sophisticated business Web application today, you're using probably between five and 10 different products. So, you get the five to 10 products, five to 10 languages, and it's horrendous. We talk about push technology and rich content and client-side Java. One of the nicest things people like about our product is it's a self-contained environment. You don't have to go outside the product [to build applications]. This gets back to what I was saying - in the Web, it's possible that the development tool becomes a feature of a platform rather than a separate category. If you think about it, in a three-tier environment, the development tools have to be very intimate with the environment.

There are Java development tools [from firms like] Symantec and Borland... There are a lot of people writing Java and they're going to use those tools. In fact some of them will use those tools to enhance ours. And that's fine. We view those as complimentary, as we view HTML editors as complimentary. If you want to talk about Borland they are focused on JavaScript. Now, we get into a whole discussion of the dynamic HTML, Jscript, JavaScript Active Server page approach to the world versus a pure Java approach to the world. In fact I think that's the fundamental direction issue in the marketplace today. Everybody agrees that the only way businesses are going to put out significant applications is if they have really good database capabilities, if they have really strong user interface client interactivity and you can't do that with state-of-the-art HTML today. So there's two paths trying to make that better - one path is saying 'Java, Java, Java'. And that's certainly what SilverStream subscribes to. The other path is saying 'let's enhance HTML a little bit to make it dynamic. Then let's have a separate scripting language to HTML that can manipulate the properties in realtime of HTML. And then let's have another language on the server that that can talk to. And then lets have a component model, whatever we do will have to be consistent with JavaBeans because Java beans are going to be an accepted standard, but then let's try to push our own proprietary standard, and by the way, in terms of dynamic HTML, we'll do our own flavor of it. We'' propose it to the W3 consortium and see who wins'

So you can see, it's trying to say 'I'm going to take an application server I'm going to take logic on the client; I'm going to tie these two together. In this case they're taking some of the existing technologies and try to evolve them forward and wire them together as opposed to saying here's a single universal language, a single universal interface and have a complete Java solution. These are two fundamental directions in the marketplace. And we believe that no one who started with a blank sheet of paper with designing anything other than a Java solution because they think it's clean, it's consistent and everything else is wiring together pieces.

Do you worry that SilverStream has bet so heavily on Java's success?

I would say that we could not have built what we built any sooner. And I'm a great believer in shooting ahead of the target in the marketplace because one thing you can always count on in our industry is that technology will improve, it will get faster, it will get more robust - unless it gets dropped by the marketplace. Now with over half a million Java developers, there's no chance Java's going to get dropped by the marketplace. And industry politics is one reason Java will be around as well because there are a lot of people backing Java as a Microsoft alternative.

There's an enormous impetus behind Java. And I'd always rather be early with the technology than late. And I can give you a very good example - when we started development with PowerBuilder, Windows 3.0 was not in beta yet. And very early in the development cycle we jumped on Windows 3.0 beta. At the time, there were people who said to us, 'you're taking a big risk going with a beta product and with Windows which is immature -go with a nice solid operating system called OS/2.' We decided to take a chance with Windows, and we experienced a lot of pain in the first year or two dealing with an immature technology. But it's pretty clear that if we had gone with OS/2, that would have been the wrong choice. Today, I think everyone can see that. So you pick the right technology and even if it's a little early, you know it's going to be there.

When we talk to customers - admittedly when they come to us they have already bought in to Java - one thing that dominates their thinking about Java [is] they all say that they think the Java platform will be there over the next 12 months. And I see a lot of specifics in terms of things happening over the next several months that make me believe that: the arrival of JITs, the arrival of run-time code optimizers, the arrival of native compilers for servers, the general fine tuning - the new Microsoft JVM offers a significant advancement. So Java is progressing very rapidly. Would we like Java to be three months ahead of where it is today? Yes. But is that a reason to pass on the Java platform which we believe is the right technology? No.

Agreed that it's going to get better. But in the initial going, what if its speed back and forth to the server is slower than people anticipate? There could be a backlash.

It's possible, but let's talk about performance a little bit. First of all, when you talk about an application server typically you're spending at least 80% of your response time in the database, the operating system, and the network, which means you're spending 10 to 20% in your code. If you make the execution of your code 10 times faster, you don't get 10 times faster processing, you get 20 percent maybe because you're not affecting the rest of the processing. We could put a caching algorithm into our server that reduces the number of I/Os in half and that would be more of a performance gain than Java will get through the rest of its history. So, you have to have the perspective of how much of that is absolutely an issue. Another thing to think about when you think of performance is when you're doing straight HTML, you have a round trip to the server to do anything. You have no local intelligence.

When you have Java running on the client [talking to the server], first of all, you can make heavy use of background threads because we human beings are terribly slow compared to computers, and there are lot of things as I user I do that can be processed locally without a round trip to the server. Now, there again, you can make Java much, much faster, and nothing you do can compare to avoiding round trips to the server. So, I think it comes down to is there's no question that Java will get faster and we'd love for it to get faster - don't get me wrong, we expect it will get faster - but it's not as clear as some people think. We're talking about three-tier business applications that involve significant database access, that involve significant network overhead, and Java isn't as big a percentage of the puzzle as people might think.

The Web has been acid for some big players. One thinks of Klaus Besier of SAP and Jim Manzi of Lotus. Where do you step carefully as you go into this market?

Again, the Web is such a broad term, and there are lots of different markets. "Ecommerce," for example, we're not in that market. We're not in the market for setting standards for data interchange. We're not in the Web market based on advertising. Our business model is actually a pretty traditional corporate IT type of business model. People want to build applications and we give them a development and deployment environment to do so. It's a new technology but in a lot of ways, it's a traditional business model. The problems of business have not changed since the technology was a hammer and chisel on rock. And we just use different technologies to automate them and create new opportunities with new technologies. I don't see IT organizations coming in and saying that they want application server software for free. They're saying they want it rock solid and fast. They want the fastest most cost-efficient way to get there. If the choice is 'roll your own,' or to use five or 10 different products, it's deadly.

Generally the industry was impressed with Powersoft, particularly in its marketing. What did you do right there?

You talk to venture capitalists and they'll tell you that when start-ups come in and they ask what's there business model they'll say it's the Powersoft business model, and what a lot of people mean by that is that Powersoft was one of the first corporate I/S types of companies to go with a multifaceted distribution model. There have been companies that have more of a direct sales model. And the reality is the direct sales people have the tendencies to run into conflicts with their channel. Powersoft was able to get a lot of people on board because we were good partners, because we had a multifaceted channel model. I think that was something very significant. You know, we get a lot of credit for marketing but you can only market a weak product so much. You can only market a great product so much if there's no market. Our industry has the tendency to one day wake up and say, 'we're not doing applications that way anymore, we're doing them this way' and within a 12-month period the entire industry turns over in this direction, That happened with Powersoft - people said 'no more dumb terminals, all new development is going to be done with nice graphic interface.' Powersoft understood the nature of that new form of application and articulated the message well and spread the word through a broad partner channel quickly. Here at SilverStream, you have a similar phenomenon - the industry is saying 'we're not going to build our applications the old way, we're going to develop our new applications as Web applications.' SilverStream is coming out and showing that we understand the implications of the Web applications. We're communicating the message crisply and clearly, and we're out of the gate starting out with a worldwide distribution channel to help convey the message as well as to help support and build those applications.