JBoss: The J2EE Open-Source Alternative
By Richard Karpinski, TheOpenEnterprise
Nov 20, 2002 (11:51 AM)
URL: http://www.theopenenterprise.com/story/TOE20021120S0001
There's no doubt that Java application servers are becoming commoditized: Sun is giving away versions of its app server and IBM will in effect throw one in if you buy enough supporting infrastructure. But JBoss -- both the open source technology project and the services company that supports it -- has been (along with Apache's Tomcat project) perhaps the biggest instigator of the open source, and free, J2EE movement.
Founded in March 1999, the JBoss application server is downloaded to the tune of more than 150,000 times per month. JBoss touts a recent survey by tools vendor TogetherSoft that found 43 percent of respondents used JBoss for enterprise development with the next closest competitor, BEA, at just 29 percent (note: most independent analysts rate IBM and BEA holding a lion's share of the enterprise app server market).
JBoss has strong momentum, but it hasn't come without controversy. JBoss is built as a "clean-room" implementation of Sun's J2EE specs. Until very recently, Sun actively discouraged such projects. JBoss is working to gain J2EE certification from Sun -- typically an expensive proposition -- but hasn't scored that important marketing benchmark yet.
Today, JBoss founder and president Marc Fleury says that while JBoss is 100 percent J2EE compliant, the server is actually much more than a J2EE container. He estimates that only about 30 percent of JBoss is J2EE-related; the rest is enterprise-ready middleware ready to support any size Web application deployment, Fleury said.
The proof is in the contracts. JBoss has run recent deals with Corporate Express and the government of Norway that represent some of its biggest enterprise wins yet.
We talked with Fleury about the future of JBoss, open-source J2EE, Microsoft's .NET, its Java rivals and much more. His goals are anything but small. He wants his service-oriented company -- driven by free and open JBoss software -- to become "a monopolistic but responsible provider of Web infrastructure."
Q: Give us the thirty-second history of JBoss.
Fleury: I was working at Sun on the SAP Java effort and working on the J2EE and EJB specfications. Later I left Sun and started JBoss as an open source implementation of the EJB specification. It was the first open source effort focused on EJB and it attracted a lot of talent early on. Pretty quickly, we got to JBoss 1.0 and an open source implementation of EJB that was leading-edge. Today, we have some features that our competitors still haven't copied.
The way we got to where we are today is through massive adoption of JBoss, crossing over now to mainstream IT. We did it by winning niche projects in large corporations thanks to our features and the quality of our product. IT all starts with a good product, dynamic proxies, interceptors, invokers, class loaders, all stuaff that simplifies every-day development. That's how we got in.
Q: Until very recently, Sun was openly hostile to open source Java implementations, right?
Fleury: Historically, Sun has not open-sourced J2EE. They open-sourced the servelet specification, which is a very small subset of J2EE, maybe ten percent. Recently, they've opened up and we're working with them to find a way to achieve certification of JBoss, which is a marketing check mark for a lot of people.
The J2EE spec is publicly available. We do not ship any classes from Sun, we did a clean-room implementation. What JBoss is today, there is very little that is just J2EE-specific, it is more of an enterprise-wide framework for Web applications. It's just really good middleware.
Q: Many open source products have had early enterprise success in small areas, like Linux with print and file servers, and grown from there. Has JBoss had a similar experience?
Fleury: Absolutely. Adoption was fast. Unlike most open source projects that have a high barrier of entry to adoption, JBoss by following the J2EE standard found a way into enterprise development immediately. A few years ago, we saw a lot of development on JBoss but then they'd deploy on IBM WebSphere or on BEA. Now, we're in the third generation and we see JBoss everywhere, in development and in deployment. Bit by bit we are gaining bigger and bigger projects.
In the extreme high-end, with many CPUs, we are a very strong competitor. The way people approach it, if you've got 300 CPUs, JBoss with zero dollar per CPU is a strong option. Granted, if there's a high-end deal dictated by hardware and services and the app server to boot, it's hard for us to compete. But if you look at a high-end, purel application server deal with services to boot, then we compete very favorably.
Q: Another way commercial vendors compete is via tool, server and add-on server (portal) integration. How do you compete there?
Fleury: In many cases where a company goes with JBoss, the integrated stack has very little appeal. The attraction of IDE and portal integration is actually fairly small. We have a best of breed approach. Every company uses several different IDEs and portals and what-not. When you talk to vendors, the main reason they want a customer to buy into their entire stack is that we've crippled their ability to make money from just their app server and they need to derive revenue from somewhere else.
Q: Are we coming to a point, spurred by open source, where the software industry will be driven less and less by license revenues?
Fleury: We really see a fundamental shift in how people want to spend their money on infrastructure. They see and and understand open source infrastructure. The shift that it has enabled is that companies now want to spend money on knowledge on how to use that middleware. In the open source arena, J2EE and app servers is a very service-intensive field. You asked if I believe that we'll see the end of licenses for infrastructure. The answer is yes. I also believe there's a monopolistic opportunity in open source infrastructure, just like Microsoft has a monopoly on the desktop. Free software will create a market that is much more open than that, but we see ourselves becoming a standard, used everywhere, while other application server vendors are struggline. That's our end goal, to become a monopolistic but responsible provider of Web infrastructure.
Ссылки по теме:
