Maybe i am wrong, perhaps the app server is a dying platform, i heard it straight from the product manager of JBoss recently that the world does not revolve around app servers, a seemingly innocuous comment that resonated with me for being incredibly brash considering the revenue model associated with JBoss – – I mean is there a single other product at the JBoss division that has any chance of making money?Not rules, not transactions, not development tools, not even SOA, at least not without being a JBoss AS customer; this is why i am so comfortable going to bat for the long-ridiculed product, and forever-predicted demise of the platform that changes the economics of software…there is nothing since the operating system itself that created a market for ISVs, developers, and their customers to collaborate on the emergence of a new sub-industry within software; the database has not accomplished in 20 years what the app server has in 10…so, it is with relative optimism that i look to a future of distributed platforms that run business transactions that inter-operate based on standards that the app server sets where business is heading; looking forward to something to replace app servers is often associated with the supposed demise of Java as the premier contrary language to a Microsoft world…this is not likely to happen anytime soon, either, so with that, i will turn my attention to how the world of enterprise software looks set to shake-out with some anecdotal remarks on what the vendors can do: 1. Pick a platform and go with itOracle did that yesterday with WebLogic, and they should be congratulated for just being realistic to the prospects of ongoing development on OC4J, it just doesn’t make any more sense since they have the original Java application server in their arsenal. IBM, on the other hand, has yet to make this vital decision, and instead choose to bi-furcate the WebSphere and Geronimo decision with utter nonsense when it comes to pushing customers to one code-base (Geronimo) when it comes to ‘entry-level’ development, and to another (legacy WebSphere) when it comes to deployments; they need to re-orient all future development around Geronimo a.s.a.p.2. Utilize OSS to compete with .Net Glassfish is an under-heralded success in the open source community, as it is the true standard-bearer for Sun’s software biz, more than openSolaris and MySQL; Sun needed to do it, and demonstrated why it is so important to utilize the Java community at large to create a platform. SpringSource is the second example of what OSS development can mean to creating critical mass, as they have moved in to tier 1 middleware status on the implementations of the legions of developers coding to Spring. And JBoss is the undeniable originator of the OSS middleware model, and will continue to have the leadership mantle as long as execution remains in view.3. Focus on strengths, not ubiquityThere is too much at stake, and too much competition from the most viable software vendors to ever consider going after a majority of the market as JBoss GM, Muzilla, claimed in early 2008: ubiquity is not an option for anything, whether it be Java, .Net, or a vendor’s solution. What should be done is get the most revenue, deployments, and mind-share in this interim period before the Internet becomes a truly global marketplace with transactions for e-commerce and supply chains being conducted on app servers. The examples are plentiful: Oracle sells WebLogic to its database customers; Red Hat sells JBoss on Linux deployments; Sun sells deployment capabilities on its Java hardware; IBM does the same….but it seems that only Oracle is willing to make the necessary adjustments to prioritize their product-line and core customer base over the mantra of being all things to all audiences… 4. Promote JEE.Net is not and never will be dead, it is only a matter of time before 50% of departmental applications are built from Microsoft tools, but that does not mean that 50% of the influence in the market will go to MSFT, the real money is on the back-end with automated transactional processing and business process management – – this is where Enterprise Java has an undeniable advantage for integration and scalability. JEE is the one thing that stands in the way of Visual Studio converting the entire enterprise development community and its associated customer base; calling JEE d.o.a. or ‘irrelevant’ is narrowly-focused banter, and is purely ignorant of the trend to distributed systems based on Java web services, including EJB end-points, JAX, JBI, and other associated specifications that enable the app server to inter-operate with .Net…5. Focus on deployments, not development For years, we have been hearing that developers hold the keys to all decisions, and this is still true to an extent, but developers are notoriously investigative in their downtime, and need to stay within company practices when they are on the clock at Fortune 2000 clients, as well as doing contract work for SMBs…the vendors need to provide better or any tools to help with the deployment and integration of JEE applications and components in order to make the transition from experimental development in to hardened deployment more smooth; the standard-bearer for this type of work can be found in the previous iterations of Sun’s Reference Architectures, as well as countless other best practices from tier-1 vendors…6. SOA is about re-useIntegration needs to develop along standards, and there is plenty of old code that can be incorporated in to modern Java apps through the Java web services specifications; components for business logic are the ideal, but countless opportunities exist to provide re-use at all levels of software integration…the application server is the glue that will incorporate all the functionality to make SOA useful… 7. Go for the kill on the back-endDo not let Microsoft splinter the Java market through better marketing on a one-stop shop approach, make inter-operability the core value proposition that incorporates portability of Java across platforms. The Spring team and EJB expert group need to come to an agreement on a direction forward, and the onus is on both to utilize best practices to stop the vitriol, and focus on the common enemy; this more than anything would enable Enterprise Java to take the mantle from Microsoft for critical mass development and deployment, nothing less is a success… Java