Date of this Version
Although XML was originally envisioned as a replacement for HTML on the web, to this point it has instead been used primarily as a format for on-demand interchange of data between applications and enterprises. The web is rather sparsely populated with static XML documents, but nearly every data management application today can export XML data. There is great interest in integrating such exported data across applications and administrative boundaries, and as a result, efficient techniques for integrating XML data across local- and wide-area networks are an important research focus.
In this paper, we provide an overview of the Tukwila data integration system, which is based on the first XML query engine designed specifically for processing network-bound XML data sources. In contrast to previous approaches, which must read, parse, and often store XML data before querying it, the Tukwila XML engine can return query results even as the data is streaming into the system. Tukwila features a new system architecture that extends relational query processing techniques, such as pipelining and adaptive query processing, into the XML realm. We compare the focus of the Tukwila project to that of other XML research systems, and then we present our system architecture and novel query operators, such as the x-scan operator. We conclude with a description of our current research directions in extending XML-based adaptive query processing.
Date Posted: 09 May 2005
This document has been peer reviewed.