Hadoop Based Data Processing Method and Its Application on Braille Identification

Abstract

It is not so efficient to store and manage massive data with traditional distributed database architecture. Hadoop, as an open source framework, is suitable for data distributed storage and management on a group of cheap machines. Based on the research of open source frameworks of Hadoop, a kind of massive data processing model based on Hadoop is provided in this paper, and its application on Braille identification is introduced. The paper also provides a test scheme for I/O performance of this data processing model under different numbers of data node and different copy coefficient and block size. The test results show that, the model provides high efficiency for large scale of data processing.