id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_fx34tla4enhr3a3igkgbwuqajq	Ralph R. LeVan	Parallel Text Searching on a Beowulf Cluster using SRW	2005		.htm	text/html	3176	251	74	OCLC provides a search service (FirstSearch [3]) that makes available both browser and Z39.50 [4] access to the WorldCat database. We chose to partition our large database into a number of smaller databases and use federated searching techniques to search across the databases with a Beowulf cluster [6] as our hardware platform. With that scheme, we only need to sort the item records within the partition and can have the sorting done on the application nodes, instead of having to collect all the results on the root node and doing the sorting there. OCLC Research has developed open source database software that consists of two modules: the search engine (Gwen [10]) and the database building software (Pears [11]). While running the SRW client test, we noticed that the root node for the cluster, where the client was running, was red on the Ganglia page, and all the search nodes were blue.	./cache/work_fx34tla4enhr3a3igkgbwuqajq.htm	./txt/work_fx34tla4enhr3a3igkgbwuqajq.txt