Question

I’m working on a shortest path a* algorithm in java with a mysql db. I’m executing the following SQL Query approx 300 times in the program to find route connections from a database of 10,000 bus connections. It takes approx 6-7 seconds to execute the query 300 times. Any suggestions on how I can speed this up or any ideas on a different method i can use ? Thanks

private HashMap<Coordinate,Node> closedNodes;
private PriorityQueue<Node> openNodes;

..
private List<Coordinate> calculatePath() 
{
    //While there are nodes in the open list
    while (!openNodes.isEmpty()) 
    {
        //Get the node with the lowest gVal+hVal
        Node node = openNodes.poll();
        //Add it to the closed list
        closedNodes.put(node);
        //If it is not the goal node
        if (!node.equals(goal)) 
        {   
            //Get all the neighbours and Create neighbour node
            List<Node> neighbours = helper.getNeighbours(node, goal);
            //For each neighbour
            for (Node neighbourNode : neighbours) 
            {
                //Check if the neighbour is in the list of open nodes
                boolean isInOpen = checkOpenNodes(neighbourNode);
                //If it is not in the open nodes and not in the closed nodes
                if ((!closedNodes.containsKey(neighbourNode))&& (!isInOpen)) 
                {
                    //Add it to the list of open nodes
                    openNodes.add(neighbourNode);
                }
            }
        }
        else 
        {
            // We found the path
            path = backTrackPath(node);
            break;              
        }
    }
    return path;

/**
 * Gets the list of valid Nodes that are possible to travel to from <b>Node</b>
 * @param stopNode Node to find neighbours for
 * @param goal End Node
 * @return list of neighbour Nodes
 */
public ArrayList<Node> getNeighbours(Node stopNode, Node goal) 
{
    ArrayList<Node> neighbours = new ArrayList<Node>(); 
    Node neighbourNode;     
    //get neighbours connected to stop  
        try {
            ResultSet rs = stmt.executeQuery("select To_Station_id, To_Station_routeID, To_Station_stopID," +
                    "To_Station_lat, To_Station_lng, Time from connections  where Connections.From_Station_stopID ="
                    +stopNode.getCoord().getStopID()+" ORDER BY Connections.Time");

            rs = stmt.getResultSet();
            while (rs.next()) {
                int id = rs.getInt("To_Station_id");
                String routeID = rs.getString("To_Station_routeID");
                String stopID = rs.getString("To_Station_stopID");
                String stopName = rs.getString("To_Station_stopName");
                Double lat = rs.getDouble("To_Station_lat");
                Double lng = rs.getDouble("To_Station_lng");
                int time = rs.getInt("Time");
                neighbourNode = new Node(id, routeID, stopID, stopName, lat, lng);
                neighbourNode.prev = stopNode;
                neighbourNode.gVal = stopNode.gVal + time;
                neighbourNode.hVal = heuristic.calculateHeuristic(neighbourNode, goal);
                neighbours.add(neighbourNode);
            }
        }
    catch (SQLException e) {
        e.printStackTrace();
    }
    return neighbours;
}

Answer 1

In general, if your query is slow and expensive, try caching the results somewhere, so on the next lookup it will be retrieved quickly from cache. So you would (expensively) compute connection between point A and B, store the entire result set in other (temporary=cache) table in the database with a defined lifetime, so for the next X hours/days (or until the routes change) you can retrieve route from A to B from this cache table.