Real-time analytics processing with MapReduce

摘要

Many real-time analytical applications over massive data streams were performed by usually introducing a specific stream processing core. In general, these SPCs were not popularly applied to enterprises same as MapReduce, even if now real-time analytics applications are taken into attention more and more. For reversing this tide, we developed a new analytics system. Our system modified the stock Hadoop's MapReduce programming model and execution framework, and used Chord model as temporary data, Cassandra as its persistent storage. With our system, we can develop data stream processing application with the familiar MapReduce programming model.