I would like to make catchments for many thousands of points using Network Analyst. The number of points is too big for one "Solve," so I am iterating through each point, solving and exporting the shapefile

I used Model Builder to start:

enter image description here

It works, but after a few thousand catchments it begins to slow down. Watching the geoprocessing window, I see "Copy Features" is the culprit: after each iteration it seems to hang on it longer and longer. The add location/solve seem to operate at the same speed. I tried adding a delete "in_memory" to the model but it didn't help. I would like to know what could be causing the "Copy Features" to drag, and if it is possible to fix it.

With model builder it is hard to figure out what is going on. I would like to switch to python but am stymied there as well (see this question)

  • Is it possible that your copy features is copying ever increasingly larger datasets?
    – Hornbydd
    Commented Jun 14, 2016 at 22:35
  • Each output is a shapefile with one feature. I don't know if it is cycling through an increasing chunk of memory to find that feature -- I tried to find a method for clear/remove/delete polygons in the SA, but nothing worked. I could remake the SA with every iteration but that seems clunky
    – woot
    Commented Jun 15, 2016 at 6:01

1 Answer 1


I've come across similar issues with mass automation generating thousands of files.

An issue is that you are filling a folder with thousands of datasets which all have one feature, this would have an impact.

I would consider either writing them to a file geodatabase or combine them in some manner.

  • You could turn your model into a sub-model and pass out a collection of files to a master model with a merge tool. This approach works if you process your data in batches.
  • You could try to append the output of the network solve into a single featureclass.

I think it would be smarter and far more efficient to collect your output into a single dataset rather than generate thousands of individual datasets.

  • In another project I had similar problems with Appending tables: the first few thousand were fast (.4 seconds) but it would slow -- after 5-6000 it would be 2 seconds each and eventually it failed at 15K. As for GDB, maybe I need to experiment again. When testing, I found it much slower to write a file to GDB as opposed to a shapefile in the short run. But maybe it won't suffer from the same slow down
    – woot
    Commented Jun 15, 2016 at 9:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.