Towards Providing Complete Knowledge in Constraint-based Database Caching

Database caching accelerates access to frequently requested data of a remote database (backend) by keeping subsets of records close to applications. Thereby, the constraint-based approach primarily uses constraints to determine what predicate extensions are complete and which queries can be answered. Equipping the backend with so-called complete knowledge about the caches’ configurations enables it to make sufficient predictions about the contents stored in the caches. Furthermore, complete knowledge obviates the unnecessarily high communication costs caused by the currently supported simpler structural knowledge during update propagation. In addition, with complete knowledge it becomes possible to prepare the loading of contents on backend side. This paper describes the basic requirements and proposes an index structure that enables complete knowledge on the backend.