From Schema Evolution
Jump to: navigation, search


AIMS is an efficient incarnation of Prima, which supports the same features of management and querying of transaction-database systems under schema evolution. The original PRIMA is based on XML DB that executes XQuery queries on XML data, which is not very efficient at the current state of art. In AIMS, we instead employ RDBMS-based query execution engine, to improve the performance. We also address the problem of temporal coalesce for the fragmented history by schema changes. For more detail check the paper at SIGMOD 2010 <bibref f="defbib.bib">Sigmod2010</bibref> The main investigators are:

Hyun J. Moon (contact author): [1]

Carlo A. Curino: [2]

Carlo Zaniolo: [3]


The problem of archiving and querying the history of a database is made more complex by the fact that, along with the database content, the database schema also evolves with time. Indeed, archival quality can only be guaranteed by storing past database contents using the schema versions under which they were originally created. This causes major usability and scalability problems in preservation,retrieval and querying of databases with intense evolution histories, i.e., hundreds of schema versions.This scenarios are common in web information systems and scientific databases that frequently accumulate that many versions in just few years.

Our system, Archival Information Management System (AIMS), solves this usability issue by letting users write queries against a chosen schema version and then performing for the users the rewriting and execution of queries on all the appropriate schema versions. AIMS achieves scalability by using (i) and advanced storage strategy based on relational technology and attribute-level-timestamping of the history of the database content, (ii) suitable temporal indexing and clustering techniques, and (iii) novel temporal query optimizations. In particular, with AIMS we introduce a novel technique called CoalNesT that achieves unprecedented performance when temporal coalescing tuples fragmented by schema changes.Extensive experiments show that the performance and scalability thus achieved greatly exceeds those obtained by previous approaches. The AIMS technology is easily deployed by plugging into existing DBMS replication technologies, leading to very low overhead;moreover, by decoupling logical and physical layer provides multiple query interfaces, from the basic archive&query features considered in the upcoming SQL standards, to the much richer XML XQuery capabilities proposed by temporal database researchers.


<bibreferences />

Personal tools