XML multi-core query optimization based on task preemption and data partition

Abstract

In XML query optimization, most algorithms still use the traditional serial mode, so they can hardly take full advantage of multicore resources. This paper proposes data partition for the XML database so that the new approach reaches load balance between different cores. Each thread processes the sub-regional data independently to reduce synchronization and communication overhead between cores. This paper also discusses the usage of task preemption in multi-core querying. According to experiments, our strategy requires less time consumption and gains better workload balance than both NBP and SBP. © Springer International Publishing 2014.