An Optimal Cache for a Federated Database System

摘要

Federated database systems allow users to query different autonomousdatabases with a single request. The answer to those requests mustbe found on the underlying databases. This answering process can beimproved if some data are cached within the federated databasesystem. The article presents an approach that allows the definitionof an optimal cache for a federated database system according to aset of parameters. We show the types of objects to be cached, thecost model used to decide which ones are worth caching and the methodto find the optimal set of objects to cache. Moreover, this approachcontinuously updates the set of parameter values and periodicallyredefines the optimal cache in order to reflect changes in the userrequirements or in the implementation features of the underlyingdatabases. The article also presents how cached data can be used toanswer a user query. Furthermore, the advantages of using a KnowledgeRepresentation System based on Description Logics in order to definean optimal cache for a federated database system are shown throughthe paper.