Data Cube Computing Method Research For Real-time OLAP

OLAP is the most important decision support analysis tool in data warehouse.The traditional OLAP designment is based on historical data, it emphasizes on theoffline batch computing which restricts real-time analysis. In today’s rapidlychanging business society, policymakers need to grasp the fleeting businessopportunities, and request for real-time OLAP analysis to help them make decisionsin time. Real-time OLAP asks for fast response ability while data warehouse isupdating. In order to realize the real-time OLAP, data cube is an effective solution,precomputs all the possible query conditions user may raise. When user submitsqueries, data cube will return the expected result directly, without the need forcomplex online aggregate computing. While the data volume of data warehouse isvery huge, and keeps a highly growth trend, and the corresponding data cube sizewill be dozens or even hundreds times bigger than the original data volume. Datacube computation is a data intensive and computational intensive task, differentaggregate calculations and storage have very high requirements for both time andspace.For the requirements of real-time OLAP, the idea is as follows:First of all, data cube is expected to solve the response problem. In this paper,the computation method of data cube is studied, in order to reduce the timecomplexity and space complexity.Secondly, in order to solve the real-time updating problem, we mainly studiesthe incremental updating supported data cube model and incremental updatingmethod.Finally, because the data cube is unavailable during the updating period, so wefurther studied the response strategy during the updating period, to meet therequirement of real-time OLAP.The contributions of this paper are shown below.First, through a lot of contrast experiment, we found that, in order to achievereal-time OLAP, data cube compression is very necessary, and effective data cubecompression method will greatly cut space complexity as well as shorten theaggregate calculation time and incremental maintenance time. We use the extendiblemultidimensional array as the data cube organization model, and eliminates thetotally redundant cells, at the same time, the HOMD based physical storage modelcompress the data cube size again, so we have greatly reduced the data cube spacecomplexity.Then, for the first time in this paper, we proposed the MOLAP based data cubecompression method SC-Cube, and put forward the corresponding data cube establishment and incremental updating algorithm.Finally, we further studies the SC-Cube based online aggregate method, inorder to meet the requirements of the real-time online analysis queries.