| Sorting is a core operation in database management system(DBMS). Energy-efficientsorting has become a research focus in energy-efficient DBMS. Existing XML sortingalgorithms are almost restricted to serialization without taking advantages of multi-corecomputing; In addition, existing XML sorting algorithms induce lots of disk accesses andrandom I/Os. With all these issues in mind, we aim to solve these problems and propose a newenergy-efficient XML sorting algorithm.In this research, we design the XML sorting algorithm from two perspectives: performanceand energy-efficiency.1.For performance-oriented, we propose a hierarchical data sorting algorithm called XPSort.XPSort reduces the maintenance-cost of in-memory XML tree by managing memory datacarefully and avoiding sub-tree matching; XPSort effectively reduces disk IOs through datacompression and single temporary-file mechanism; XPSort exploits multi-core to parallelize tasksorting and task generating.2.For energy-efficient oriented, after analysing the difference between performance sortingalgorithm and energy-efficient sorting algorithm, we put forth an energy-efficient XML sortingalgorithm called EEXPSort by adapting XPSort.Experiments show that XPSort has better performance than HERMES, which is the fastestXML sorting algorithm currently, and that EEXPSort has better energy-efficiency thanHERMES and XPSort. |