Alovera: A Fast Stream Processing System for Large-Scale Data

Abstract

Growing of data volume poses challenges to data processing system. In this paper, Alovera, a fast stream processing system for large-scale data is presented. By using columnar data layout and stream processing, it is capable of pipelining data processing efficiently. It can process part of data instead of waiting for all data to be ready for the next operation. Thus, it can reduce the query time dramatically. Experimental results indicate significant performance improvement in a variety of tasks. In the experiments, we also evaluate our methods with different systems including HadoopDB and Hive. The extensive experiments confirm efficiency and better performance of our system.