Due　to　the　unlimited　amount　of　information　available　on　the　Web,　it　is　a　burdensome　task　for　users　to　navigate　the　ever-increasing　Internet.　Therefore,　Web　search　engines　become　necessary　tools,　supporting　information　searching　and　retrieval.　However,　it　is　a　challenge　to　build　all　efficient　system　to　search　around　the　highly　dynamic　World　Wide　Web　(WWW).　The　aim　of　this　paper　is　to　develop　a　high-performance　Web　search　system.　In　this　paper.　we　propose　the　architecture　of　a　distributed　and　multi-threaded　search　system,　and　develop　this　system　in　Java　on　IBM　ABLE　platform.　With　the　assistance　of　various　techniques　concerning　crawling,　XML,　and　database,　the　paper　makes　special　emphasis　oil　the　design　and　implementation　of　Web　crawling　model　and　pages　extraction　model.