Design And Implementation Of XML Update System Based On Region Encoding

As a new markup language, XML has become the standard for representing and exchanging data via network. Since its unique advantages such as self-description, and platform-independence, more and more applications use XML to store, exchange and publish information, which brings a great challenge to XML storage solution.With powerful and reliable data management capabilities relational database is increasingly widely used to store XML data.Currently the performance issue of query and update xml data based on RDBMS has become a research hot point.In order to use existing RDBMS it requires us to efficiently map xml data into relational table.However, with no aware of underlying tree-shaped data RDBMS can't make full use of encoding information, which affects its query and update performance. Region-based encoding maps xml data into index structure that facilitates the management.More importantly, Staircase Join are aware of tree-shaped data and can effectively reduce redundant operation for loop.Experiments show compared with other approaches the use of region-based encoding can significantly improve the performance of XPath and XQuery.However, because of structural update problem in region-based encoding nearly half of the total nodes need to be changed for update operation.Therefore, it is very meaningful to design and implement a high-efficient region-based xml update system.This paper designs and implements a prototype system called XUpdate2SQL. It consists of two major components:The first part designs the rules to map XUpdate to SQL and uses loop-lifted approach to convert loop in XUpdate to join in SQL to solve the differences between XUpdate and SQL.Then according to those rules it transforms XUpdate AST (Abstract Syntax Tree) to SQL AST with BURG tool to complete compilation process. The target language of internal compiler is SQL which follows SQL92 standard and supported by popular RDBMS so it has good versatility.The second part designs BUS(Batch Update Strategy) by introducing incremental BTree, not only reduce the number of the nodes needed to be updated but also save the time spent on reconstruction index.Tests show BUS has greatly enhanced XML update performance. BUS maintains the integrity of the internal structure of the database so it is independent of specific RDBMS and has good portability. In addition, for ordinary users who are unfamiliar with XUpdate language this system provides GUI and fast menu to automatically generate XUpdate statement and enhances ease of use of the system.Tests verify XUpdate2SQL prototype system maintains efficient query performance, while improving the XML update performance, system's overall performance is improved. Versatility, efficiency, portability and ease of use constitute the biggest characteristics and innovation of this paper.Finally, we give a conclusion of the advantages as well as drawbacks of XUpdate2SQL prototype system, which will guide our research work in the future.