Several of my Hosted Feature Layers have been increasing in size. Has anyone else had similar experiences? How do you manage this?
I am the sole publisher of this data that is hosted in AGOL, and no other members of my organization edit or modify these files.
For example:
A file's normal size is 15mb. I have modified this file, however, the total feature count, field count, attributes etc, remain similar, no major changes besides a slight increase or decrease in total records. My method of modifying the data is to add it to a map in ArcGIS Pro, delete all features, then append features. The file size remains the same for hours to days, then increases dramatically to 70mb or even more. I had one go from 30mb to 280mb AND back down to 30mb with no intervention or input, right in front of mine and Esri's eyes as we screen shared!
I realize that enabling sync or editing can affect the file sizes, however the initial jump in size is erratic and not always directly related to toggling settings. As in, it does not always jump immediately when I toggle editing or sync, it might jump a few hours later or the next day.
Yesterday, a file I have not modified in at least several days, doubled in size with no discernible triggers. I toggled editing on and off and the file size went back down to the expected size. Now, this morning, it is back up to double what it should be with no changes overnight.
I opened a ticket and conversed with Esri for weeks back in February/March, and they were not able to give concrete answers as to what exactly affects size and how to predict changes. This was their response.
While I could find some documentation on size increases on sync-enabled layers, I haven't been able to find any related to editing. However, after speaking with colleagues, it is expected for the size of a layer to increase after enabling sync and editing, as we have both seen.
It's difficult to speak to the amount a layer will increase in size, since this is dependent on the schema of the data, how many replicas may be created, the geometry of the features, etc. However, we saw similar magnitude increases (10x) in our datasets. Just something to be aware of when enabling those features.
I think this is unacceptable. In order to be able to budget for credit consumption, file settings and their affect on size needs to be transparent and predictable, not erratic and spontaneous.
I have relatively small datasets, what about organizations that have large data stores? Their credit budgets must be incredibly hard to pin down when their file sizes are all over the place.
