Large Overlay and Spaghetti and Meatballs Questions

Discussion created by csny490 on Apr 13, 2011
I have some questions pertaining to running large overlays in ArcGIS and examining overlapping features. Maybe someone from ESRI might want to share some corporate secrets?

1) When running an overlay, there is a little executable that fires off called TestGPram.exe. This is used as part of the ArcGIS tiling scheme used to chop up large overlay tasks. My question: Assuming everyone has at least 2 GB of RAM available, what are the conditions that cause a tile to be subdivided? Is it simply the number of vertices in the input FC(s) per tile - Or is it more complex? The reason I ask is that sometimes I will run a large overlay using the default tools (union for example) and the thing will sit there for hours happily subdiving tiles, create hundreds of tiles, and after 6 hours, finally finish. However, using the same input layers, I can create a four cell (2 x 2) fishnet polygon FC, set the gp.extent to one fishnet poly at a time, and run the union (one onion per fishnet poly) and be done with the union in just 10 minutes. What gives? What causes the large overlay tilling thing to go crazy creating hundreds of tiles for 6 hours when just four tiles will do the job and the thing takes only 10 minutes?!?

2. When running a "Spaghetti and Meatballs" analysis (that makes me hungry, so I'll call it an "overlapping polygon or polyline analysis"), there are at least a few tools that will "shatter" the geometry of overlapping features (aka "planarization"). For polygons: Union or FeatureToPolygon (intersect too I think), and for lines: Intersect or FeatureToLines. My question is: What is the "best" method to use for shattering polygons and/or lines? Are there other/better way of doing this that I am not aware of? For polygons, ESRI seems to like using the FeatureToPolygon tool instead of the Union tool... What is the reasoning? In my experience it is not faster, actually quite the contrary.