|
INTEGRATION WATCH: Not So Extreme Programming
By Andrew Binstock
 May 15, 2004 — Over the course of years, I have discussed many times the fate of various J2EE implementations—the early lead by BEA, the astonishing charge into first place by IBM, the lamentable decline of Sun, and the on-again, off-again push by database players such as Oracle and Sybase. In the mix were occasional mentions of other vendors, such as Allaire/Macromedia’s fine JRun product and other lesser-known platforms. Rarely, however, have I mentioned JBoss. The reason, quite frankly, is that for much of that period, JBoss was just another open-source project.
During the past year, however, as open source has become widely accepted in IT, JBoss has gone commercial and is increasingly showing up on my radar. Private clients now mention that after BEA and IBM, JBoss is the J2EE stack they hear of most frequently. The JBoss project was morphed into a company run by its founder, Marc Fleury. His business model is to offer the J2EE stack and various add-ons at no cost under the lesser GPL license (LGPL) and then charge for support and documentation.
As the product caught on, Fleury did several things right—the most important of which was hiring proven industry managers to set corporate direction. He scooped up Bob Bickel, who previously transformed a little-known consulting company that had a Java server into HP’s erstwhile middleware division. Fleury then went out and got venture capital, as reported in this newspaper in the March 15 issue (“JBoss Goes Corporate for Credibility,” page 1).
Along the way, JBoss has become a favorite solution for OEMs that need a Java stack. JBoss is OEM’d by Apple, Borland and Compuware, among others.
For end-user sites, the temptation presented by JBoss is even greater: The product is free. The documentation and support, though, require payment. The first question prospective sites tend to ask is whether the JBoss product can compete with the enterprise features of WebSphere and WebLogic.
JBoss lacks some of the sophisticated features of clustered servers, including advanced caching mechanisms, but the company is actively working on this and other capabilities. In fact, it has just released JBoss Cache, one of the first implementations of JCache, a Java Community Process-mired proposal for caching on J2EE. Due to its large infusion of VC cash, JBoss is likely to eventually have all the enterprise features of its BEA and IBM counterparts. I have every expectation that JBoss will eventually earn the No. 3 slot in the market behind the two market leaders.
But the path will not be easy, despite JBoss’ merits. The article in SD Times I referred to earlier rightly focused on the deficiencies of the company’s business model. If support is principally what it’s selling, scalability is a serious challenge, and the needed staff is implacably expensive unless offshoring can be worked out. Either way, it’s a model that is not conducive to significant growth or great profitability. Even OEM deals will not greatly improve the picture, that is, if there are many such deals left to be made.
Incremental growth has risks as it lets competitors catch up, and JBoss has competitors. Among open-source J2EE stacks, there are the Geronimo project, under the auspices of the Apache Software Foundation, and a larger project from ObjectWeb, a European consortium of mostly French companies and institutions. ObjectWeb’s JOnAS is mature and constantly being revved. There also are numerous Java containers available in the open-source community, the most famous of which, Tomcat from the Apache Software Foundation, is highly regarded.
While these companies compete with JBoss on products, a new start-up competes on services. The Core Developers Network (www.coredevelopers.com) was formed by a group of ex-JBoss employees. It provides training, documentation and support for a wide variety of Java technologies, including JBoss.
I believe JBoss will continue to grow despite these adverse market conditions, because it is driven by professional managers and it was the first of these start-ups to get VC funding, which enabled it to build the corporate infrastructure necessary for subsequent growth. It will play the part of MySQL in its relation to PostgreSQL.
Due to better funding and a full company behind it, MySQL is the open-source DBMS to bet on, even though a portion of the market will remain loyally attached to PostgreSQL. Unfortunately for JBoss, it does not enjoy MySQL’s dual licensing scheme by which the Swedish firm can obtain payment for its open-source product when customers cannot meet (or abide) stringent licensing provisions. Nonetheless, J2EE stacks require lots of expertise, and so JBoss might be able to thread the needle.
If JBoss permanently takes over third place in the J2EE market, it will validate the model that most open-source companies hope will work: one that is built on free software and paid services. But until it reaches that point, we won’t know if the model can be made to work.
Andrew Binstock is the principal analyst at Pacific Data Works LLC
|
|
Advertisement
Click here for a complete listing of Integration Watch Columns
Click here to see a complete Column Archive. |
Back to Top
|