Research And Implementation On MySQL Storage Security

With the era of knowledge economy, we become increasingly dependent on information. Information security is the core of database security, so database encryption to sensitive information is a theoretical and practical significance of the research topic. Data is encrypted on the outer layer of the DBMS usually, but this method is lack of security and performance. Transparent encryption implementation on the database storage layer, the decryption process is transparent to users. Storage layer of the DBMS can be an effective solution to the shortcomings of the data encryption on outer layer, so it has a great research value.First MySQL architecture is introduced, then the InnoDB storage engine source code. The core algorithm of the InnoDB storage engine that binary search and B+ tree is analyzed and modified: decryption function is called after read data, then the decrypted data compared with the query; encryption function is called ,then the encrypted data stored into the location determined by the binary search and B+ tree search; the search algorithm is called to find data after operations of delete and update, then you need to decrypt the ciphertext, modify the transaction field and rollback pointer field, mark the data for deletion or update the field , finally encrypt to storage.This paper analyzes the mechanism of storage engines page split, the left split of the data page, the data split in middle of the page and the right of the page. In the process of page split, records will be split into the new page according to different circumstances which is determined by the improved insertion algorithm, and you need to do some encryption and decryption processing to those records.This paper analyzes the principle of the index of the storage engine. InnoDB storage engine use merge sorting algorithm. Data is inserted into the new page after sorted and encrypted which is constructed from the clustered index.Finally, we test the system function, construct test cases, simulate the three case of page spilt , and the system is able to correctly insert and query.