MapReduce performance in MongoDB sharded collections

Abstract

In the modern era of computing and countless of online services that gather and serve huge data around the world, processing and analyzing Big Data has rapidly developed into an area of its own. In this paper, we focus on the MapReduce programming model and associated implementation for processing and analyzing large datasets in a NoSQL database such as MongoDB. Furthermore, we analyze the performance of MapReduce in sharded collections with huge dataset and we measure how the execution time scales when the number of shards increases. As a result, we try to explain when MapReduce is an appropriate processing technique in MongoDB and also to give some measures and alternatives to take when MapReduce is used.