| 摘要 |
<p>Disclosed is a database and Hadoop hybrid platform-oriented OLAP query processing method. The method uses a multi-copy fault-tolerance mechanism of a Hadoop as a basis, and comprises: when performing OLAP query processing, executing same on a master work copy first, and recording a query processing result in an aggregation result table of a local database; and when a work node generates a failure, looking for node information about a fault-tolerance copy corresponding to the master work copy through a namenode, and invoking a MapReduce task to complete an OLAP query processing task on the fault-tolerance copy. The present invention combines the database technology with the Hadoop technology, and combines the storage performance of a database with the high expandability and high availability of a Hadoop in a manner of dual storage engines and dual query processing engines; and uses a master-slave type copy management mechanism to integrate database query processing with MapReduce query processing in a loosely-coupled manner, thereby ensuring both high query processing performance and high fault-tolerance performance.</p> |
