The age of Big Data is coming. With data sets increasing, traditional relational database management system has bad performance, and it has been unable to provide an efficient service to real-time system. NoSQL database become the better choice to handle massive data sets. Therefore, how to transform the existing data form SQL database to NoSQL database becoming an important issue. Apache HBase is the most popular NoSQL database and widely been used in many areas. Thus we experimented with MySQL and HBase in this paper and sum up three different types of SQL-to-HBase schema transformation method. We also optimize the performance of HBase by detecting the correlation between Tables in the SQL database.