A metadata access strategy of learning resources based on HDFS

According to the existing project of visual cloud computing mobile learning platform and considering the performance bottlenecks of NameNode in the Hadoop Distributed File System (HDFS), e add a metadata server (MDS) to share part of the metadata access pressure when the load of NameNode is too large, to optimize the performance of the overall system. This paper details the improved metadata access strategy; compares the differences of overall system performances between the improved one and the original one through performance tests. The conclusion is that the improved metadata access strategy can improve the overall system performance when the NameNode is under great pressure.