Transparent Query Rewriting For Unstructured Data Stored In RDBMS

The cost of data management is instantly increasing by years with the scale of dataexpanding. For effective management for unstructured data, NoSQL non-relationaldatabase is developing rapidly. However, in real applications, the type of data is divers. Itis hardly dealing with data. So it is a challenge for relational database dealing withstructured data and non-relational database only dealing with unstructured data. Now,although using relational database to manage unstructured data and non-relationaldatabase to deal with structured data，the performances of these are not good. They spendmuch more time to import the data into the database, and make the data fit the way ofstorage.The system for Key/Value data stored in relational database makes relational databaseas its storage. The original tables were split into small tables which have only twodimension based on attributes of value. Then the content of small tables was exchangedinto int by Hash. This method for storing unstructured data not only makes up for thewaste space which is produced by splitting the original tables, but also it is helpful for theperformance of the query. Query rewriting is based on the schema of unstructured datastored in relational database. To realize query rewriting, the important part is to realizequery rewriting of DDL and DML. So, using SQL to query unstructured data will berealized.The results of the experiments show that the performance of the storage forunstructured data in non-relational data storage system is good. For the query forunstructured data, the experiments show the system is good at decomposing the statementsubmitted by users, and generating the corrected sub-query statements.