Network alarm information analysis - A Hive-based OLAP method

For network alarm information analysis in the next generation network existing such problems as single dimension, low efficiency, memory overflow or crash, this paper proposes a method for online analytical processing based on Hive. The method first uses HBase to preprocess real-time alarm data, mapping data into Hive, then constructs an N-D cube model. After proving the scientific and operability by cube calculation, is the construction of a star model, as well as customizing dimensions, each dimension with user-defined hierarchies. Finally HQL is used to realize rollup and cube. This new method is available for both situations whether inter domain devices exist or not. It can not only upload large quantities of data, improving performance of upload and analysis, but also discover knowledge of location and meaningful information hidden in massive warnings from multiple dimensions, providing decision support for network management personnel.