Pretty much as in the subject. Snow is an R package for doing mapping of computations onto processes in one or more servers that's simple to use, and requires little configuration. Organizations sometimes use Hadoop and Spark to manage large clusters of processors. Is there a way for snow to coexist with Spark? My personal experience is using snow's SOCK interface (as opposed to MPI or PVM or NWS).
More on snow is available at http://homepage.stat.uiowa.edu/~luke/R/cluster/cluster.html -- Jan Galkowski Senior System Software Engineer, II. Akamai Technologies, Cambridge, MA Custom Analytics Research & Development -- The magic of statistics cannot put actual numbers where there are none. -- Paul Holland --