Would it be possible to use KNIME in some sort of distributed parallel processing?
I understand KNIME supports multiple CPU processing in one computer, but how about 2 or 3 computers in a network that contribute resources to one common task or a model?
If this is possible, how can it be inplemented (like for example do I need to install some additional drivers etc.).
You are right, this is a great extension. And the good news is that it is already available.
KNIME Cluster Execution allows running a complete workflow or single nodes on an SGE Cluster. You gain significant performance and partner nodes that normally do not support cluster execution can be run parallel as well.
You have to keep in mind though that only calculations that are done row by row are suited for Cluster Execution. If your node needs access to all rows then splitting will lead to wrong results.
You can find details on Cluster Execution at: http://www.knime.com/products/knime-cluster-execution
If you want more information or a demo let me know.