Knowledge　discovery　in　datasets　integrated　into　Grids　is　a　challenging　research　task.　These　large　datasets　are　being　collected　and　accumulated　across　a　wide　variety　of　fields,　at　a　dramatic　pace.　They　are　often　heterogeneous　and　geographically　distributed　and　globally　used　by　large　user　communities.　In　this　paper,　we　proposed　the　scenario　that　deploys　a　grid-based　parallel　data　mining　service　in　the　Globus　4　Framework　and　hosts　within　Jakarta-Tomcat　as　a　container　in　detail,　using　OGSA-DAI　for　access　to　Grid　Data　Source.　The　work　presented　in　this　paper　reveals　specific　requirement　for　applying　grid-based　data　mining　in　the　service-oriented　grid　environment.