Load balancing parallelizing XML query processing based on shared cache chip multi-processor (CMP)

Chip multi-processor (CMP) could support more than two threads to execute simultaneously, and some executing units are owned by each core. Based on CMP, this paper proposes a novel and complete optimization framework on parallelism for XML database multithreaded query processing that strives for maximum resource utilization. Firstly, a set of algorithms for constructing the parallel sub-query plans and partitioning XML document by parallel sub-query plans are proposed. Furthermore, in order to reduce cache access conflict and address the imbalance of threads’ workload, we refine the granularity of partitioned XML document and balance the workload assignment by executive pairs for the unit. Finally, by building the execution plan tree of sub-query plans constructed, the partial solution produced by parallel sub-query plans are merged into final solution. Our theoretical analysis and empirical evaluation show that the proposed algorithm could impressively improve in the performance of XML query processing.