| With the development of network technology, the amount of data generated by each dataplatform constantly increasing, data warehouse architecture made more rigorous requirements.Although nowadays there are a lot of massive data processing program, such as: Exadata, Hadoop,etc, but the cost is too expensive Exadata makes prohibitive for many companies, long constructionperiod Hadoop technology costs are high for those non-computer technology as the core of theenterprise is also very difficult. How to use the relatively inexpensive cost of building a reliableperformance for many enterprise data warehouse system ardent aspirations.This paper studies the design and implementation of ORACLE based distributed datawarehouse. The idea of using a distributed database design and construction Weibo user behaviorfingerprint analysis data warehouse. Include: a summary of the data warehouse design,fragmentation design of distributed data warehouse, the overall design of a distributed datawarehouse. In the summary of the design, the proposed algorithm using the association rules to dealwith the properties of microblogging users, complete the fingerprint of user behavior statistics; inslice design, combined with the data presented data warehouse processing logic vertical slicealgorithm based on graph theory; overall In the design, focusing on the global data dictionarydesign control and monitoring modules distributed database system design; last part of the analysisresults show.The system leverages the distributed components ORACLE through horizontal and verticaltwo slices simplifies data processing logic and memory pressure single node, while completingtransparent data distribution and query by passive global data dictionary. The entire systemstructure is not complicated, and highly scalable, has some practical value. |
