Automated Data Versioning with a Storage Engine Middleware

While MySQL offers a wide variety of extensions on user level, the storage engine extends the ABI at the lowest database level. Traditional engines, such as MyISAM, InnoDB, PBXT, deliver a data store for MySQL with different transaction isolation methods and performance. With the appropriate methods, the Revision Engine ABI can add very low level functionality to MySQL in a pluggable way.

With these methods the engine can extend the basic data store paradigm to a data versioning store. The versioning store does not claim to be a new persistent storage engine but is more of a middleware between the MySQL core and existing persistent storage engines. Hence only minimal code is used to implement a most reliable (in terms of bugs in the code) software stack. Moreover the revision paradigm can compete with existing techniques (such as triggers on tables or revision logic within the client application). Compared to the traditional versioning techniques, the revision engine uses low level MySQL interfaces. In this manner it is able to perform more efficiently on DML statements than if the versioning is performed in the client software or with triggers.

In this presentation we will show the benefits of extending MySQL on a storage engine level in terms of data security and DML performance. Furthermore we will illustrate common relational based data versioning and how they can be implemented.