摘要
MapReduce是一种分布式的并行编程模式,它可以实现大型数据集的并行运算。Lucene是Apache下的搜索引擎开发包,当索引文件不断增大时,Lucene搜索便会出现瓶颈问题。通过利用MapReduce的思想,按城市划分策略将大量并发的搜索请求映射到对应的分布式服务器中进行Map操作,再结合Lucene,从对应索引服务器中查询后利用Reduce操作返回最终结果。实验结果表明,这不仅解决了大数据量查询的瓶颈问题,还将系统效率提高了66.7%。
MapReduee is a distributed parallelized programming model. It can implement the processing and generating large data sets. Lucene is a Search Engine API under Apache. When the index file growing, the Lucene Search performance is a bottleneck. Based on the MapReduce, this system maps the parallelized search request to the cluster server for Mapping operation. It is mapped by dividing the index file by city strategy. And then the Map Function get the search results with the lucene. The results will be returned to the user by Reduce Function. According to the experimental results, this design does not only resolve the paralleized search bottleneck, but also improves the performance by 66.7%.
出处
《计算机系统应用》
2012年第2期249-251,224,共4页
Computer Systems & Applications
