Multiversion Repeatable Read Isolation Level – Theory and Practice

Concurrency control is the activity of synchronizing database operations by concurrently executing transactions on a shared database. We examine the problem of concurrency control when the database supports multiple versions of the data. Multiversion concurrency control is used in order to improve the level of achievable concurrency. The goal is to produce an execution that has the same effect as a serial one. We use the multiversion concurrency control theory to analyze histories produced by multiversion concurrency control methods. We show that using traditional repeatable read isolation level is inadequate and provide a new isolation level definition for multiversion repeatable read. We show that the new isolation level captures the essence of repeatable read isolation level in multiversioning systems.