基于Hadoop云计算平台提供的Pipes工具和多层次k-way图划分方法,实现了并行化的OSLOM(POSLOM),该算法能有效地减少计算所消耗的时间.实验表明,相比于OSLOM,POSLOM具有良好的可扩展性,可以获得近乎线性的加速比,同时精度损失可以控制在5%以内.
By using the Pipes tool and multilevel graph k-way partitioning method based on cloud computing platform,the parallelization of OSLOM(POSLOM) was presented,which can effectively reduce the time required for calculation.Experimental results show that compared with OSLOM,POSLOM has a very good scalability,and it can get speedup ratio almost linear,while the accuracy loss can be controlled within 5%.