基于GNP算法的分布式爬虫调度策略

GNP-based scheduling strategy for distributed crawling

针对分布式搜索引擎的任务调度及负载均衡问题,提出了基于GNP算法的分布式爬虫调度策略和负载均衡的方法。利用网络距离预估取代大规模的网络距离测量,不仅提高了系统的响应速度,还减少了系统对广域网造成的压力。通过在广域网上部署爬虫节点,构建分布式搜索引擎,应用该调度策略进行实验,验证了系统性能有较大提高。

In order to solve task scheduling and load balancing problems of distributed search engines, this paper proposed a GNP-hased scheduling strategy for distributed crawling and a load balancing method. Adopted internet distance estimating mechanism as a replacement for large-scale network distance measurement, which not only improved response time of the system, but also reduced WAN pressure caused by the system. Through deploying crawling nodes at WANs, built a distributed search engine, and implemented several scheduling strategies. The online experiment shows great improvement in system＇ s performance.