Panta Rhei: Database Evolution

Опубликовано 6 сентября 2016, 18:31

Panta Rhei---everything is in a state of flux. Evolution impacts Information Systems (IS) at many different levels, we focus on the problem of evolving the data management core of an IS. We present an ongoing activity of collecting and analyzing a large number of evolution histories of real-life IS. The results of this analysis confirm the initial hunch about the severity of the problem, and provide the foundations of a benchmark for tools supporting schema evolution. This work served as a motivation and invaluable test-bed for the design and development of the Panta Rhei Framework, which provides support for: (i) assisted and predictable evolution design, (ii) data migration scripts generation, (iii) automatic query (andupdate) rewriting, (iv) transaction-time archiving and querying of databases under schema evolution, and (v) automaticdocumentation of the metadata histories. The presentation will focus on two components of the Panta RheiFramework: PRISM and PRIMA. PRISM achieves (i)-(iii) by harnessing recent theoretical advances in schema mapping and query rewriting in a concrete design that provides practitioners with the needed support for graceful schema evolution.PRIMA supports complex temporal queries on transaction time databases under schema evolution.By decoupling the logical and physical layers PRIMA provides the users with a rich temporal query interface (based on an XML temporal data model) while building on a reliable relational storage engine.The complexity introduced by schema evolution is completely masked to the user exploiting the rewriting technology we developed in PRISM. Temporal-specific query optimizations and a performance- oriented architecture complete the system.