Even though NetKernel is shrouded with new edge terms like 'Transreptor' and 'Fragmentor', it is quite a practical piece of software. Listen to Jon's 'Innovator' podcast with Pete at 1060 (yeah Tony, your an Innovator too) for a audible view into the NetKernel architecture.
Since I have adopted NK at my company, it has started to fill blanks in much of our architecture.
We use it for two main purposes. Content aggregation and systems integration.
With content aggregation, NK pulls data from one or more sources, possibly applies some rules, and feeds the results to the presentation side systems. One example of this is serving Nutch indexed search results of our content. Another is where we query a RDBMS for relevant documents and then pull together the referenced complex xml fragments from a back-end system, over HTTP, and return merged results.
And with systems integration, we have built a JMS transport that delegates incoming messages to NK services that were built in a transport agnostic way, thus allowing those services to be tested over HTTP with no additional complexity in the service. This particular application is a jewel in our infrastructure, and has replaced all components of our vendor provided ESB (except for CBR services) implementation (see downfall for relevant bits).
To be honest, we may choose to replace the CBR services with NK as well since NK plus mod-e4x has given us massive productivity gains. That said, I need to update the library at some point. Pete and Tony have put many hours in to fixing various bugs in Rhino and XMLBeans that I'm likely the sole recipient of.
Leave a comment