How to Store and Query XML Data

Due to the fact XML is a universal data-exchange format, all the questions concerning storage and querying XML documents have been lately of great popularity. Query languages, called XPath and XQuery have been developed and standardized in order to query XML data not only in XML files, but anything existing as XML, including databases. XML databases running XPath and XQuery are generally classified as native and relational. The native ones are specialized engines evaluating queries on XML documents, while the relational are built on top of existing database engines. The intention of this paper is to discuss some of the main problems and research trends from the bread research field devoted to XML storage and querying. The exposition will start with a description of XML Data Management Systems, go through the query languages XPath and XQuery, examine in details the aspects in the development of XML query processing and end up with the latest proposed algorithms for efficient query processing. XML Data Management Systems Existing architectures of XML database management systems will be presented and commented with respect to storing and querying opportunities of XML data. 1) Storage of XML documents in relational systems To store XML data in relational systems has been popular for the last several years and as well implemented by few commercial DBMS. The reason is that it is based on adaptation and reuse of relational technology. However, there are significant differences among the existing methods on how to save and hence query XML data in RDBMS. Shredding XML documents into relations XML documents are transformed into atomic values and after that stored as relations in relational tables. XQueries are translated to SQL queries to be evaluated by the RDBMS query processor. Several different shredding techniques and query capabilities has been proposed [14, 15]. The advantage of this method is that it does not require big modifications of the existing database engine. XML, stored as unparsed text XML data is stored in VARCHAR or LOB (large object) columns of relational tables. XML data is queried by a XQuery processor external to the database and invoked as a user-defined function. This approach is used by commercial systems offering XML support (MD2, MS SQL). This solution is relatively simple but the entire XML document usually has to be first loaded into the memory in order to be processed. Hybrid XML-relational databases XML documents are stored on disk pages in tree structures matching the XML data model [7, 9, 16]. Hence no mapping is needed between XML and relational structures. The native XML storage is complemented with XML indexes. It is provided XQuery and SQL support.