I need to create a data model for routing data between points. For example, path between two cities which will include the two start and destination cities, way points (like smaller towns along the way), total distance between the two cities (distances between routing points do not need to be modelled though it would be good if that can be done too).
My current idea is like below.
- Create one table containing the cities:
- Create a many-to-many table
pathwith 4 columns:
city.city_id), calculated column
distanceto store the total distance for the route.
- Create a table to store route points called
townsimilar to the
- Create a man-to-many table
townwhich will have three columns:
route_positionwhich will be a value 1-n which will indicate the position of the route point in the actual path. For example, first route point will have number 1, second 2 and so on.
- Do you guys think this is a feasible model? Is there a better approach to this?
- In the
pathtable, is it better to use a
MD5hash or create a multiple-column index on
To give some context, I'm currently using a graph database to store this data and have a user-defined function that uses A* search to find the shortest path between points. Problem is as the graph gets denser, this computation becomes too slow. So, I think a lookup-based approach would be faster.