Geronimo and the Java application server market

Sean Gallagher was responsible for a series of interesting articles this week about Java application servers that I read while catching up on things today (there were more but that'll do).

What's missing in the series is an external perspective. While hinting around about potential doom and gloom for vendors who sell app servers, he didn't actually talk to the vendors. More importantly, there wasn't any attempt to talk to real world users of any stripe.

Application servers (of all flavors) are a large part of an enabling platform for everything from business applications to web services to pure web applications. I'd argue that too few use the rest of stack (the world is not just Java, there's typically a great unix layer sitting there too). To go along with the large number of applications comes an even more eclectic bunch of organizations and individuals. There is no 'one size fits all' characterization that can be applied.

In The Geronimo Effect and Will Geronimo Stick a Fork in Java? it seems to be argued that Geronimo will cause a sea change of sorts in the market. Maybe, but I wouldn't count on it happening anytime soon. A major shift (the major shift?) was already brought about by the JBoss folks when they proved a significant level of interest and involvement in an open source application server (standards and Sun's bless be damned).

Geronimo is going to be a good thing for all who appreciate choices and diversity. It was probably a requirement for the Apache folks to add a J2EE app server given their strong Java support.

At the same time, open source doesn't appear to have significantly impacted the fortunes of Sun, IBM and perhaps even BEA. Many organizations want products that work out of the box (and some want them to run on Windows), have binders full of documentation and army's of internal and external support structures.

Sun bundles Sun ONE with practically everything, has an interesting licensing model and doesn't seem to want anything more from the product than to help push Java technology. IBM has done some wonderful things to tie Websphere into Eclipse and make it a brain dead simple and a very compelling marketing message, they're in this for long haul and don't appear ready to go anywhere. BEA Systems is a whole lot harder to peg, but they helped push the whole idea of application servers to the mainstream and remain a fertile source for ideas; it would be a shame to see them go away.

What's really bizarre about the articles is the apparent concept that change is bad, especially when the source is Marc Fleury's cant on licensing: On the Issue of Business Friendliness.

I like JBoss a lot, and pushed to make the move away from closed systems like WebLogic (not all groups played along). I love being able to develop, deploy and run on any Java enabled platform, fix bugs in any part of the application stack and control my world as needed. That doesn't make me blind. While ranting about the short comings and changes in the Axis API's, there is no discussion about numerous zigs and zags on the JBoss platform. We're talking about major changes (often with significant developer impact) in every single major version change. Hey, that's just technology if you're a JBoss customer but a mean perversion of BSD licensing if IBM does it. Please... grow up.

I wish the JBoss group well, as they walk an interesting line between a traditional OSS and being a service oriented corporation. I do think someone needs to stuff a handkerchief in Fluery's mouth and pay more attention to service organizations like Jabber, Inc. and MySQL AB.