A study of different XML parsing techniques in a multicore environment
详细信息
文摘
Extensible Markup Language (XML) is a known language encoding used in semistructured documents. XML has been widely applied in databases, web servers, architectures, and messaging systems. However, inferior performance of processing XML documents has been identified. There are many efforts to address XML's inferior performance. Multi-core architectures, which utilize more than one core for higher processing throughput as opposed to increasing clock speed have gained significant attention in computing. Lu, et al. proposed a novel parallel parsing approach. Partitioning XML documents into small chunks and placing them in a multi-core environment achieved the desired improved performance. The shape of an XML document is not known until parallel parsing is complete. Because of XML's unbalanced structure there is a varying work-load among individual cores. In this paper, we combine this approach with a work stealing mechanism and show that this can improve performance in a multi-core environment.