Model-Driven Parallel Green Computing of Big Data Systems (Green-BiD)

Green computing is the study and practice of environmentally sustainable computing or IT. The term covers a broad set of techniques including design, implementation and use of systems efficiently and effectively with minimal or no impact on the environment. An important goal of green computing is to maximize energy efficiency during the product's lifetime. Green computing is important for a broad range of computing systems, ranging from handheld mobile systems to large-scale data centers.

In this project we will provide a model-driven approach to analyze, model, and select feasible mappings of parallel algorithms to a parallel computing platform to reduce the power consumption of data centers. In the approach we will provide the steps for defining models of the computing platform and the parallel algorithm. Based on the analysis of the algorithm and the computing platform feasible mappings will be generated. This will be supported by a corresponding toolset that builds on a predefined metamodel. Using model-to-model and model-to-text transformations we will provide a solution to the code generation and portability problems. The evaluation of the approach will be based on real world case study and simulation platforms.