OpenMP and MPI were preferred for parallel programming in the modes of shared memory and distributed memory, but now research has been focused on hybrid programming using both MPI and OpenMP and Apache Spark programming, which targets fast big data analysis.
Commonly, however, configuring a cluster environment requires more time than programming, and moreover, the difference between MPI-OpenMP of a parallel system using C language and Apache Spark of a distributed system using Java is one of the challenges to anyone who are not familiar with configuring Linux cluster as well as deploying Apache Hadoop stack.
Kim Ryo Chol, a section head at the Faculty of Information Science and Technology, proposed an integrated platform in which MPI-OpenMP and Spark parallel programming are both possible.
Virtual machines needed for parallel computation are provided by the private cloud based on OpenStack IaaS.
Applying the proposed method, he built an integrated platform for parallel programming, thus reducing the time for cluster environment configuration.
For more information, please refer to his paper “An Integrated Platform Architecture for MPI-OpenMP and Spark Parallel Programming on Cloud Environment” in “Proceedings of KUTIC-2025”.