Buy Vs Build with open source alternatives

Recently, we evaluated a few commercial products to solve a particular business problem. Though, the products had very impressive features, the licensing costs were daunting. Moreover, I felt that the features offered by the commercial vendors could be built in house using a combination of proven open source technologies. I mentioned what I had in my mind to my management and they readily approved and encouraged me to develop a POC within two weeks and give a demo.

The open source softwares I picked up are highly proven and famous - JBoss Rules/Drools, Quartz scheduling engine, Jackrabbit CMS. These have been nicely integrated and I knocked out the POC within a week. Many thanks to the committers on these wonderful softwares. Ofcourse, there was a bit of pain as there was no sufficient documentation and tutorials available, I had to hack around the API and source code, but over all it was a very exciting experience. Now my current task is to port the POC to the server side and provide hooks to access the functionality via web services. Hmmm, this brings in some of the routine enterprise chaos :-)

The challenging task is to deploy the Jackrabbit CMS on the Weblogic. The example shipped with source code talks about Model 2 way of deployment on weblogic as an JCA adapter. Pretty interesting! Note that JCA resources can be accessed only EJBs, servlets and not from a main method to unit test. As I look more at the situation, many concerns are getting unearthed. Little did I realize that the amount of entropy in a software system is proportional to the breadth of the system. This gets maginified further if different technologies are weaved (I desist from using the word "integrated") together and at the same time the adjecctives of the system are to be taken care of viz. flexibility, extensibility, etc and more importantly performance. Can we call this nice pabulum? Indeed! I also have the satisfaction of saving a few hundred thousand dollars for my company!!