Enhancing XML Data Parsing and Querying Performance on Multi-Core Architectures

Abstract

With the increasing number of computer applications, the eXtensible Markup Language (XML) has become ubiquitous and is being utilized broadly on the web for exchanging and storing information. XML files of large size take more space because of redundancy, and when parsed on machines degrade the execution. This paper proposes a novel approach aimed at enhancing the performance of parsing XML files. The proposed approach initially divides the XML file into several parts concurrently and then processes it to ensure well-formedness. Each part is subsequently encoded to save memory space, efficient parsing, and querying. Multiple parse trees are then generated in parallel in the next phase, prior to querying required data from the parse trees. The parallel parsing and efficient querying make our approach perform better than other parsing approaches. The results obtained on processing XML files of diverse sizes show an increase in performance by 22.50%, 19.41%, and 30.25% over the well-known DOM, SAX, and StAX parsing approaches, respectively.