Need collaboration that does not consume so many credits

05-13-2021 03:25 PM
New Contributor III


  • I need to supply a vendor with 24 separate feature services in ArcGIS Online.
  • The data for these services must be stored in our enterprise environment, and edits to the source data need to get to AGO at regular intervals.
  • Grouped layers are not supported by the vendor.
  • Our firewall only allows edits to go out, not into our enterprise.

Initial solution: 

  • I made a collaboration between our AGO and Enterprise which worked perfectly.
  • I set the collaboration to send copies once an hour rather than use references because copying created an AGO rest url that the vendor could consume (reference collaboration creates an enterprise URL which cannot be consumed by vendor since they are not behind our firewall). 


  • The hosted feature services gobble up too many credits in AGO. 
  • The increase in storage does not seem to be related to requests hitting the services or the hourly syncing (see screenshot where storage credits go up at irregular intervals).




Thank you in advance!




0 Kudos
1 Reply
Occasional Contributor III

How many megabytes of features are you storing? Its easy to estimate your costs (many assumptions including credit costs are pro-rated and not including any credits built into your subscription):

Megabytes of Feature Storage / 10MB * 2.4 credits * 0.1 $perCredit * 12months = $perYear

1MB = $0.29

100MB = $28.80

1000MB = $288.00

Once you know your costs you could check out other commercial providers who can host features and make services available to try and find a better deal. But that would have to be by WFS and not Esri REST API I suppose.

If you are getting up into the 10's of gigabytes then the dedicated feature store options make more sense pricewise, as they include 1 terabyte of feature storage in the price.

There are other ways to reduce the cost as well. Do all the attribute fields need to be included? Are the length of text fields optimised? Could text strings be replaced by codes? Possible to optimise the data to reduce the number of vertices etc.?