Even though your data science platform might support parallel job execution, it is up to the individual creating the job to make sure it is written in such a way that it can truly be executed in parallel across a large cluster.
Most machine learning algorithms are not designed with parallelism in mind. They have been created to run on relatively small datasets in single-node environments—and to get some models to run in parallel can be extraordinarily complex.
TIBCO® Data Science – Team Studio software takes the burden of redesigning common algorithms away from the data scientist. This solution brief explains how it's done.