Research Of Distributed Search Engine Based On Solr

Due to the fast development of intellectual of transfer system and increment in the amount of instrument like customs,the number of records of cars passing by increases in a fast speed,forming massive data.In a custom system,users often need to do exact or fuzzy search for number plates and the system should be as reliable,fast and sufficient as possible.In a situation with billions of records and high concurrency,traditional search engine cannot guarantee real-time response,causing low query performance and bad user experience.In order to improve it,a way using search application server to create index for the cloud storage database comes into being.In the setting of massive data of transportation,this paper come up with a distributed real-time search model based on Solr,specially designed for massive data and high concurrent requests.Considering time information are always included in transportation records and time is often used in sorting or searching conditions,this paper optimizes search application server in indexing,merging index,search index and sorting to provide efficient retrieval service.So that search application server can be improved the ability to response requests in transportation industry.Because the indexing of Solr based on master-slave replication clusters,this paper raises an algorithm to assign master or slave nodes to optimize the cluster's load balancing.When some server is down in the cluster,the algorithm can automatically reassign master and slave nodes,and recover when servers come back on line,so that every server can be used reasonably.