mapreduce技术原理解释（Mapreduce技术原理解释）.docVIP

mapreduce技术原理解释（Mapreduce技术原理解释） In order to realize the operation of the HBase data sheet must first understand its principle: The first part: the Hbase principle HBASE is an open source project based on Hadoop, an implementation of Google is BigTable. BigTable is the most simple view of a large table, table according to demand attributes can be dynamically added, and no association between the query table and the table but demand. BigTable is a distributed database system Google designed for large-scale structured data storage and distribution of the internal management, is one of the Google cloud infrastructure, has good scalability, can be in the thousands of cheap petabyte level data stored on the server. It does not provide the full relational data model, the data provides only a simple model, which makes the client can dynamically control the data layout and format. BigTable is a Dabiao, this form in fact and traditional database we imagine the table or some difference. This table is a sparse multi-dimensional mapping table, sorted distributed permanently stored on the hard disk. The mapping table for keywords, list of keywords and timestamp as index, each value is an array of bytes without explanation. Line key can be any string, each row key under the read and write operation is atomic, no matter how many read and write for different columns. BigTable by keyword dictionary order to maintain data, a dynamic table can be divided into a plurality of consecutive rows, continuous line called Tablet, it is a basic unit of data distribution and load balancing. BigTable the list of keywords into groups, one for each column, column family is the basic BigTable access control unit. Usually, the data stored in the same column with the same type of family. Before creating a list of keywords to store the data, you must first create a column family. The column number can not be too much in the family in a table, the number of columns is not restricted. The BigTable table