MapReduce-based warehouse systems: A survey

Today´s world is of data explosion, the size of data sets is increasing continuously with the huge amount of volume, where various data sets being managed and examined are called “Big data”. The analysis of Big data is growing rapidly in the industry for business intelligence making traditional warehousing solution prohibitively expensive instead of conventional database systems which have difficulty to manage Big data. MapReduce is a computing paradigm that has not only gained a lot of response now days from research and industry, but also recognized as a more effective tool for large-scale data analysis. The MapReduce framework and its open-source implementation Hadoop provide an accessible and fault-tolerant structure for large scale data analysis and it has been successfully built in social network websites and major Web service providers. This paper overviews MapReduce-based data placement structure for Big data analysis, namely row store, column-store, hybrid store, RCFile, Mastiff and ORC File.