Selective 2-versioning for concurrency control in data warehousing systems: S2V

Abstract

To speed up on-line analytical processing, data warehousing systems materialize views derived from several operational databases. When the operational databases happen to change, the materialized views get stale. To enhance the freshness of the materialized views, we proposed a new concurrency control scheme, which enables a number of update transactions to execute concurrently with analytical transactions. When an update transaction attempts to modify the data that is using by any analytical transactions, this scheme immediately allows the update transaction to create a new transient version for the data. Due to this selective versioning, analytical transactions could continuously access fresh data.