I've recently been given a list of 'routes' from a client. The list simply comprised two sets of coordinates, for the start and end point of each route. In this dataset there are around 350 routes. I firstly plotted the start and end points as two separate feature classes.
My client wants to see the routes through the road network, rather than a simple crow-flies pairing. Using Network Analyst within ArcGIS 10.3, the best option seems to be to run a 'Closest Facility' calculation. I treat the start points of my routes as facilities, and the end points as incidents, and sort/differentiate both datasets using the route name field provided by the client.
As some of these routes overlap, and don't necessarily represent the closest pairing of points, I'm not actually looking for the closest facility at all, but this seems to be the only network analyst tool within the extension that produces actual routes through a network as an output. To get round this I've ramped up the number of 'Facilities to find' to 50.
Having run the calculation, the tool produces a series of polylines. I then have to export this as a new featureclass, and split the 'name' field (the result field). I then look for pairs where the first half of the 'name' field is equal to the second half, as the name field simply records the route names of the start and end points. However having run this tool I've found that some start points are never routed to their corresponding end points if they are particularly far away and (presumably) there are fifty closer end points, or incidents, for the software to analyse.
I suppose the obvious option is to cut my dataset down into a number of subsets. Having tried to increase the number of incidents for the tool to find my computer promptly ran out of memory (to paraphrase the error message).