Views on the Industry

glassfish v. jboss?

In the emerging drama that is enterprise Java middleware platforms, I would like to provide some background as to the probable key development in the market for 2006: the introduction of a legitimate competitor to JBoss, in the form of Glassfish. I have written extensively about both, and have even covered the also-ran app server platforms that vie for JBoss' attention, but really only Glassfish appears positioned to be an alternative. Let's start by giving you punters a link to the command central of the Glassfish community, the Aquarium:

http://blogs.sun.com/roller/page/theaquarium

Now, lets talk about what would be a true competition, if either party dares to compete with the other. Marc sounded off first last summer around Java One, when Glassfish was introduced:

http://jboss.org/jbossBlog/blog/mfleury/2005/06/06/SUN_open_sources_XYZ_who_cares.txt

And since he has done nothing but admire the software moves of Sun lately, it would seem that he and his advisors are coming to terms with the next OSS JEE platform, and that it would be wise for JBoss to welcome the entrant rather than pick a fight. Why is this? Well that is the topic of this entry, lets get to it...

When JBoss first ran in to app servers, they were running in to a highly fragmented market of proprietary solutions, masked as standards-based under the guise of the J2EE Compatibility Kit. With a mantra, a cause, and a legit solution, it did not take long for JBoss to take advantage of the grand dysfunction of the no-portability Java enterprise market. Only WebSphere and WebLogic also benefitted from the JBoss scenario because they out-marketed everyone else, but in particular, incorrectly convinced customers that the app server market was gradually and definitively turning in to a natural duopolistic market. For every Oracle, there is DB2; for every Windows, there is Java; for every Netscape, there is, er, IE, and on and on. However, what Sun proved (I'll take a bow now for Sun ONE Application Server 7) is that standards matter, and provide choice, but only when implemented purely. By 2002, JBoss had eliminated the existing low-end, and IBM and BEA had developed proprietary high-end functionality for every last feature they could think of (Development, Portal, Persistence, Integration, etc.). What Sun saved with the AS7 release was a viable low-end, and layed the ground work for Glassfish, two generations later.

In conjunction with the release of AS7, Oracle picked up Orion, Red Hat invested in JOnAS, and Apache inched towards Geronimo, leaving IBM and BEA to wonder how exactly they were going to make any money. Meanwhile, JBoss was devouring enterprise customers, as well as, and potentially more importantly, all ISVs and SIs in the Java middleware marketplace. It is circa 2005 that you can mark the moment when the oxygen evaporated in the high-end, proprietary app server market, leaving IBM to buy Gluecode (read: Geronimo), and BEA to introduce AquaLogic. That leaves us in 2006 seemingly without a real fight to be had. JBoss is taking WebLogic customers, IBM is re-tooling WebSphere customers, JOnAS is outfitting Red Hat customers, and Oracle, well, we don't know exactly what Oracle is doing with Fusion, but as this blog/ author has consistently stated for 1/5 years, it is only a matter of time that WebLogic finds a home. If you are paying attention, there is one remaining competitive platform un-accounted for: Glassfish.

Glassfish and JBoss will be battling for the future of Solaris customers, with the ancilliary prize of best OSS program also in contention for Windows server deployments. In other words, outside of Linux, Glassfish and JBoss will be competing for app server dominance. The somewhat ironic reality is that these two likely competitors are natural allies, for some forseeable future, and it is all in the name of JEE5/ejb3 portability. As long as Top Link does not get bastardized to be non-compatible with Hibernate, we are probably on the cusp of seeing a true, cross-platform enterprise Java apps market built around the Glassfish and JBoss shared mission, of standards-based implementations on the platform. JBoss should welcome this with open arms, as an opportunity to drain the swamp of all WebSphere customers not making a move to Geronimo/Linux, all WebLogic customers not porting to Fusion, and all legacy, non-.Net customers running on Windows. Meanwhile, with the continued deft leadership of MB and PP, Glassfish will eliminate WebLogic from the Sun GSO price-list, enabling portability across customers, and opening up the Java Economy to components, exposed as web services, inter-operable across platforms. With all that is at stake, Marc would be crazy not to take this side-agreement, even if it means he cannot be an immediate monopoly of app servers.

With JBI-enabled ESBs bringing SOA to the IT shops of large, as well as mid-sized businesses around the global economy, Java apps become the de facto server-side implementation for discerning customers, leaving Sun happy with their unsigned JBoss collaboration. There will be .Net, I at least plan for it. But I also plan for a Glassfish and JBoss marketplace that finally implements the vision of countless individuals who built Enterprise Java, not as a proprietary solution for the benefit of Oracle and BEA to extend their customer lock-in, but as a tool for Java developers to take back control from the software vendors, and in turn, give that control back to customers, which is what will lead to larger IT budgets, more extensive implementations, and better economic fortune for all of those invested in the Java Economy. As has been said many times in this column, an independent JBoss transforms WebLogic from winner to also-ran, and a truly competitive Glassfish transforms JBoss in to a global brand on multiple paltforms. Marc, give this some thought, and make your competitive moves with this goal in mind.

As for PP, your victory is nearly at hand...

posted by douglas dooley @ 11:11 AM