Towards an Ontology-Based Semantic Approach to Tuning Parameters to Improve Hadoop Application Performance

Hadoop MapReduce assists companies and researchers to deal with processing large volumes of data. Hadoop has a lot of configuration parameters that must be tuned in order to obtain a better application performance. However, the best tuning of the parameters is not easily obtained by inexperienced users. Therefore, it is necessary to create environments that promote and motivate information sharing and knowledge dissemination. In addition, it is important that all acquired knowledge be organized to be reused faster, easily and efficiently whenever necessary. This paper proposes an ontology-based semantic approach to tuning parameters to improve Hadoop application performance. The approach integrates techniques from machine learning, semantic search and ontologies. Keywords—Hadoop MapReduce; Hadoop performance; ontology; parameters turning