MARS: A System for Publishing XML from Mixed and Redundant Storage Alin Deutsch and Val Tannen   View Paper (PDF)   Return to XML Matching & Storage (Session A5) Abstract We present a system for publishing as XML data from mixed (relational+XML) propri- etary storage, while supporting redundancy in storage for tuning purposes. The cor- respondence between public and proprietary schemas is given by a combination of LAV- and GAV-style views expressed in XQuery. XML and relational integrity constraints are also taken into consideration. Starting with client XQueries formulated against the pub- lic schema the system achieves the combined e ect of rewriting-with-views, composition- with-views and query minimization under in- tegrity constraints to obtain optimal reformu- lations against the proprietary schema. The paper focuses on the engineering and the ex- perimental evaluation of the MARS system. ©2004 Association for Computing Machinery