Résumé en anglais

To prove the feasibility of an SD-DBS, we have built a prototype called SD-SQL server. The system generalizes the basic SQL server capabilities to scalable tables. We have chosen SQL server since, to the best of our knowledge, it was the only DBMS proposing the updatable distributed union-all partitioned views. SD-SQL server runs on a collection of SQL server linked nodes. For every standard SQL command under SQL server, there is an SD-SQL server command for a similar action on scalable tables or views. There are also commands specific to the management of the SD-SQL server client image or SD-SQL server client node. In this paper, we present the architecture of our prototype and its application command interface in its 2005 version. This architecture contains more features than the reference architecture in "architecture for a scalable distributed DBSs application to SQL server 2000" by Litwin et al. (2002) and "implementing SD-SQL server: a scalable distributed database system" by Litwin and Sahri (2004). We discuss the syntax and semantics of the commands. Numerous examples illustrate their actual use. We hope to convince the reader that the use of the scalable tables is in practice as simple as that of static tables, despite some limitation of our current interface. We view our current prototype as a proof-of-concept system.