Network Dataset storage

ScottBlankenbeckler

Is it better to store my network dataset within SDE, FGDB, or leave it as a shapefile?

I know that there is additional overhead when storing address locators in the FGDB and SDE but is this a similar problem with the network dataset used in network analysis?

AlexeyTereshenkov

Hi Scott,

I agree with Joe. I use file geodatabase for storing my network datasets, too. Personal geodatabase are slower than file ones for sure. And they have limit of 2GB (which is crucial quite often). With file geodatabases, you can compress them to make read-only which will make them take less place on the disk (I've seen the decrease up to 30% in size). I use compressed file geodb for storing my network datasets which are not being edited, i.e., used only solving network analysis tasks.

I do not see any point in storing network datasets in an ArcSDE geodatabase unless you are about having a multiversioned editing workflows (i.e., many users are going to edit the underlying network data sources at the same time and you need to manage this without getting your data locked). ArcSDE geodatabase has some overhead in administering as well - since you store your data in an RDBMS, you are supposed to take care of proper indexing, allocating table data storage, tuning etc. With file geodatabases life is simpler.

There is another format - SDC - which is used by Esri for delivering Esri StreetMap product. This format provides the best performance for routing yet it is read-only (Esri's in-house development).

Avoid using network datasets based on shapefiles - they are slower, edge source has limit of 2GB (a shapefile size limit) and are less optimized for data compression comparing to the file geodatabases.

So go with the file geodatabases and will be cool. 🙂