Research Of Task Partition And Cost Model In Xquery Parallel Impelmentation

With the rapid development of Internet, XML has become the standard of datadescription and exchanging. Nowadays, XML has been applied in much more fields,which leads to the amount of XML data increase rapidly. Therefore, the demand ofquery on XML data is becoming more and more widespread. The efficiency of XMLquery also was put forward higher requirements. Hence, the research about how toimprove the efficiency of XML query has become a hotspot. W3C organization issuedthe XQuery language as a formal recommendation standard to query and processXML.XQuery program becomes more and more complicated with the continuousexpansion of XML application and the increasing huge size of XML. Therefore, theefficiency of the XQuery program implementation is becoming increasingly important,and how to improve the efficiency of query and process XML by optimizing is ofgreat significance.On the other hand, accompanied by the development of multi-core technology,people attempt to improve the performance of XML query processing by using themulti-core environment. The research about parallel processing XML is more andmore. Meanwhile, XQuery, as a functional language, has inherently parallel nature todescribe the query, which provides the possibility to parallel process XML queries.However, the parallel processing researches about XML query focus on a specificpart, such as XML parsing, XML tree pattern and XML database index. There is nogood method for a whole query tasks to explore the possibility of parallel processingXML query with multi-stategies.In this paper, we propose a method to parallel process XQuery language, whichsupports multiple parallel strategies, and some specific part processed parallel.Meanwhile, we study how to describe the task in parallel implementation of XQuery.Also, we propose an algorithm to partition the query into some tasks and show how tocompute the cost of each one, which is based on the XQuery features and XML datacharacteristics. We propose the XQParti, an algorithm to divide the task based oncomputationally intense and design a dynamic cost model to compute the scale oftasks.