A similarity-aware multiversion concurrency control and updating algorithm for up-to-date snapshots of data

Real-time databases handle reading and writing of data with time constraints on transactions. Normally, data items in a real-time system have freshness requirements which need to be guaranteed, and for many transactions it is important that accessed data items origin from the same system state, which can be ensured by letting the transactions read a snapshot of the database. In this context, a snapshot at a specific time represents values on data items that were stored in the database at this time. Furthermore, similar values can be considered equal because values within given bounds do not affect the results from calculations. Previous work shows that using similarity among values of data items greatly increases the performance because there is a possibility to skip calculations. In this paper we present the MVTO-S concurrency control algorithm, which supports similarity and multiple versions of data and ensures that transactions read an up-to-date snapshot of a database. Performance evaluations show that MVTO-S increases the performance considerably compared to well-established single-version concurrency control algorithms.