PRIMA: Archiving and Querying Historical Data with Evolving Schemas

Moon, Hyun J.; Curino, Carlo A.; MyungWon, Ham; Zaniolo, Carlo
Moon, H
Curino, C
MyungWon, H
Zaniolo, C
Citations range: 
10 - 49

Schema evolution poses serious challenges in historical data management. Traditionally the archival data has been (i) either migrated under the current schema version, to ease querying, but compromising archival quality, or (ii) maintained under the original schema version in which they firstly appeared, leading to a perfect archival quality, but to a taxing query interface.
The PRIMA system, we present, achieves the best of both worlds, by archiving data under the original schema version, while automatically adapting the user temporal queries to the appropriate schema versions. The user is entitled to query the archive under a schema version of choice, letting the system to rewrite the queries to the potentially many involved schema versions. Moreover, the system offers automatic documentation of the schema history, and allows to pose temporal queries over the metadata history itself.
The proposed demonstration, highlights the system features exploiting both a synthetic-educational running example and the real-life evolution histories (schemas and data).
The selected real-life systems include, but are not limited to, the popular genomic database Ensembl and of Wikipedia, with their hundreds of schema versions.
The demonstration offers a thorough walk through the system features and an hands-on system testing phase, in which the audience is invited to interact directly with the advanced query interface of PRIMA. The conference participants will freely pose complex temporal queries over transaction-time databases subject to schema evolution, observing PRIMA rewriting and query execution capabilities.