1

For a simulation/mining on graph data I need to write thousands of data at once into a postgresql 9.5 Database with high performance. Currently I'm doing that with prepared statements and "row by row INSERTs". It works fine - but too slow. I couldn't find anything about how to write data from a collection (array, vector or ArrayList) with the Copy statement via CopyManager into a Postgresql DB Table. I would be grateful to get a simple example. How to deal with data types? CopyManager does not request for any data types.

My current Code:

package io;

import db.PostgresConnector;
import entitiesP.Edge;
import entitiesP.Graph;
import entitiesP.Node;
import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.SQLException;
import java.sql.Timestamp;



public class DbLogger {

public DbLogger(Graph g, int simNr) throws Exception {

    Connection conn = new PostgresConnector().getConnection();

    //1. Graph Data File:
        //-------------------------------------------------

            //Filename und Header:
            long timestamp = getTimeStamp();

            //Daten:
            logGraphData(conn, simNr, 
                         g.getNodes().size(), 
                         g.getEdges().size(),
                         g.getRadiusAndDiameter()[0],
                         g.getRadiusAndDiameter()[1],
                         g.getRadiusAndDiameter()[2],
                         Node.maxDegree,
                         //new SimpleDateFormat("yyyy-MM-dd HH:mm:ss").format(g.getSimStartTime()),
                         g.getSimStartTime(),
                         timestamp);



        //2. Node Data File:
        //-------------------------------------------------

                //Daten einsammeln:
                for (Node n: g.getNodes()){

                    logNodeData( conn, simNr ,
                                     n.getId(),
                                     n.getDegree(),
                                     n.getClusteringCoefficient(),
                                     timestamp);

                }


            //3. RelationData File:
            //-------------------------------------------------


                int[][] distance = g.getDistanceMatrix().clone();
                //AdMatrix dist = new AdMatrix(distance);
                //dist.printAdMatrix();
                Object[] edges = g.getEdges().toArray();

                //adjacency nodes
                int numNodes = g.getNodes().size();
                int numEdges = g.getEdges().size();

                //Edges:
                for (int i=0; i < edges.length; i++){
                    Object[] nodes =  ((Edge) edges[i]).getNodes().toArray();    //pair of nodes of an edge
                    //pairs A--B
                    logRelationData(conn, simNr  , 

                                 ((Node) nodes[0]).getId() ,
                                 ((Node) nodes[1]).getId() ,
                                 distance[((Node) nodes[0]).getId() -1][((Node) nodes[1]).getId() -1]  ,
                                 ((Edge) edges[i]).getTieStrength()  ,
                                 ((Edge) edges[i]).getNeighborhoodOverlap(((Node) nodes[0]),((Node) nodes[1]))  ,
                                 timestamp
                          );
                    //pairs B--A
                    logRelationData(conn, simNr  , 

                                 ((Node) nodes[1]).getId() ,
                                 ((Node) nodes[0]).getId() ,
                                 distance[((Node) nodes[0]).getId() -1][((Node) nodes[1]).getId() -1]  ,
                                 ((Edge) edges[i]).getTieStrength()  ,
                                 ((Edge) edges[i]).getNeighborhoodOverlap(((Node) nodes[0]),((Node) nodes[1]))  ,
                                 timestamp
                      );

                    //end edges----------------------------

                }


                //System.out.println("NumNodes: " + numNodes);

                //non-adjacent nodes
                for (int i=0; i < numNodes; i++){
                    for (int j=0; j < numNodes; j++){
                      if (distance[i][j] != 1 && i != j ){
                          logRelationData(conn, simNr,
                             (i+1),
                             (j+1),
                             (distance[i][j]==0? 999 : distance[i][j]) , //0 auf infinity setzen (999)
                             -1,
                             -1,
                             timestamp
                          );
                      }
                    }
                }

                conn.close();
        }



private static void logGraphData(Connection con, int sim_no, int numberOfNodes, int numberOfEdges,  int radius, int diameter, int effDiameter, int maxDegree, long startTime,   long timestamp) throws SQLException{

    PreparedStatement preStmt = con.prepareStatement("INSERT INTO public.GRAPH_DATA (" + "SIM_NO,NUMBER_OF_NODES,NUMBER_OF_EDGES,RADIUS,DIAMETER,EFF_DIAMETER,MAX_DEGREE,START_TIME,TIME_STAMP) VALUES (?,?,?,?,?,?,?,?,?)");
    preStmt.setInt(1, sim_no);
    preStmt.setInt(2, numberOfNodes);
    preStmt.setInt(3, numberOfEdges);
    preStmt.setInt(4, radius);
    preStmt.setInt(5, diameter);
    preStmt.setInt(6, effDiameter);
    preStmt.setInt(7, maxDegree);
    preStmt.setTimestamp(8, new Timestamp(startTime));
    preStmt.setTimestamp(9, new Timestamp(timestamp));
    preStmt.executeUpdate();
    preStmt.close();

    //con.close();

    }



private static void logNodeData(Connection con, int sim_no, int nodeId, int degree, double clusteringCoeff, long timeStamp) throws SQLException{

    PreparedStatement preStmt = con.prepareStatement("INSERT INTO public.NODE_DATA (" + "SIM_NO,NODE_ID,DEGREE,CLUSTERING_COEFFICIENT,TIME_STAMP) VALUES (?,?,?,?,?)");
    preStmt.setInt(1, sim_no);
    preStmt.setInt(2, nodeId);
    preStmt.setInt(3, degree);
    preStmt.setDouble(4, clusteringCoeff);
    preStmt.setTimestamp(5, new Timestamp(timeStamp));
    preStmt.executeUpdate();
    preStmt.close();
    //con.close();

    }


private static void logRelationData(Connection con, int sim_no, int nodeId1, int nodeId2, int distance, int tieStrength, double neighborhoodOverlap, long timeStamp) throws SQLException{

    PreparedStatement preStmt = con.prepareStatement("INSERT INTO public.RELATION_DATA (" + "SIM_NO,NODE_ID1,NODE_ID2,DISTANCE,TIE_STRENGTH,NEIGHBORHOOD_OVERLAP,TIME_STAMP) VALUES (?,?,?,?,?,?,?)");
    preStmt.setInt(1, sim_no);
    preStmt.setInt(2, nodeId1);
    preStmt.setInt(3, nodeId2);
    preStmt.setInt(4, distance);
    preStmt.setInt(5, tieStrength);
    preStmt.setDouble(6, neighborhoodOverlap);
    preStmt.setTimestamp(7, new Timestamp(timeStamp));
    preStmt.executeUpdate();
    preStmt.close();
    //con.close();

    }

public static long getTimeStamp(){
    long t = System.currentTimeMillis();
    //SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
    //String timestamp = sdf.format(t);
    return t; //timestamp;
}

}
  • Preparing the statement each time is slow. And you should batch the inserts to speed up the data. Check out http://stackoverflow.com/questions/3784197/efficient-way-to-do-batch-inserts-with-jdbc – Konrad Jun 30 '16 at 21:57
  • Thanks Konrad - I'm going to try that out. At the meantime I found out how to deal with the copyManager as a special bulk insert functionality for postgresql. – Sven Temmler Jul 01 '16 at 18:54

1 Answers1

2

I found the solution for my problem: it works very fast :)

package io;


import db.PostgresConnector;
import java.io.ByteArrayInputStream;
import java.io.IOException;
import java.sql.Connection;
import java.sql.SQLException;
import org.postgresql.copy.CopyManager;
import org.postgresql.core.BaseConnection;


public class DbLogger3 {

public static void main (String args[]) throws Exception {

        try  {
                Connection conn = new PostgresConnector().getConnection();

                CopyManager copyManager = new CopyManager((BaseConnection) conn);

                String str = new String();

                //add 5000 data sets: (important: \r\n after each record to set a line feed (postgres interpretes that as the end of each dataset) 
                for (int i=0; i<5000; i++){
                    str += "2;100;99;6.00;12.00;11.00;13;28.06.2016 00:08;28.06.2016 00:08"+"\r\n";
                }

                //transform the String into bytes:
                byte[] bytes = str.getBytes();

                //create ByteArrayInputStream object
                ByteArrayInputStream input = new ByteArrayInputStream(bytes);

                //insert into the database table (in my case: public.graph_data)
                copyManager.copyIn("COPY public.graph_data FROM STDIN WITH DELIMITER ';'", input);


        } catch (SQLException | IOException e) {
                throw new Exception(e);
        }


    }
}