Big Data Analytics on High Performance Computing Infrastructures – Technologies , Tools , Examples

Modern data analytics applications impose demandin g requirements on storage and processing facilities and are thus e volving to a new challenging class of High Performance Computing (HPC) application s. However, typical Big Data frameworks like Hadoop, Spark, etc. are nor mally not supported onthe fly on traditional HPC infrastructures. The majo r reason for this is that they are mainly based on Java technologies that do not a llow the very complex and expensive HPC hardware to be efficiently utilized; n onetheless, the sustainable performance and efficiency of the infrastructure is a very important usage factor of HPC that cannot be disrespected in the Big Data ap plication setups. In this chapter we analyse the applicability of the well-es tablished HPC technologies and tools like the Message-Passing Interface (MPI) to the implementation of data analytics application and demonstrate advantag es of this approach over the