Ballooning Feature Layer Size

[mpboyle]

We have a feature layer that over the past couple days has ballooned in size with no reasonable explanation I can find.

This feature layer contains approximately 11k line features and less than 10 fields. Every 10 minutes the feature layer has all rows deleted and replaced using a feature set and the Python API.

This morning I manually overwrote the feature layer, and the item size was around 5MB.  A few hours later, the item has ballooned in size to 250MB!?!  

The feature layer does NOT have editing enabled, no associated views, no extra indexes, does NOT have optimize layer drawing enabled, and no attachments.

Is there any reasonable explanation for why this feature layer continues to balloon in size?

[D_Atkins]

Throwing things at the wall, here:

Problem: Hosted Feature Service Size Increases Continuously

[mpboyle]

Editing is not enabled, sync not enabled.

[MobiusSnake]

I don't have an explanation, but maybe something to try.  If you're currently using deleteFeatures() or editFeatures() in the API, see whether it still happens if you switch to FeatureLayerManager.truncate().

[MichaelVolz]

Do you see any noticeable increase in AGOL credit use due to the increased hfl size?

Do you have other hfls with the same workflow that exhibit this behavior?

[mpboyle]

I was curious about the same, so I created a script that pulls the size of some of our hosted feature layers. It includes a mix of those that we update via publishing a file geodatabase, and those that we update via deleting/adding features.

The feature layers that are published from a file geodatabase do NOT show increased item size. Those that are updated using delete/add features DO show increased item size even though the number of features is NOT increasing proportional to their shown size.

We ARE seeing this ballooning reflected in our credit usage.

Below are some charts displaying what we're seeing.

Example 1: This feature layer is updated every 30 minutes using delete/add features. On Friday afternoon the size of the feature layer was about 17-18MB. It is now showing about 45MB.

The number of features and populated fields in this feature layer remains relatively consistent.

Example 2: This feature layer is updated every 2 minutes using delete/add features. On Friday afternoon the size of the feature layer was about 2MB. It is now showing about 7MB.

The number of features in this feature layer does oscillate but is likely never above 40 features.

Example 3: This feature layer is updated once a day by publishing a file geodatabase. The size remains consistent.

Example 4: Below is snapshot of our available credits near the end of the day for the past few days (the DateValue field is set to noon UTC for our time zone).  The "CreditsUsed" field is calculated from the previous day's value.

We first noticed one of our hosted feature layers ballooning in size on Friday, 12/5/2025.  Once we noticed it, I changed this feature layer to be updated via publishing a file geodatabase.

You can see the noticeable bump in credit usage on Thursday, 12/4, and Friday, 12/5.  Once the feature layer in question was changed to be updated via a file geodatabase, the credit usage remained steady on Saturday, 12/6 and Sunday, 12/7.

I would really like an explanation from someone at Esri as to why this is happening (apologies to the Esri staff I tagged, you all showed up as top collaborators).
@EmilyGeo @RussRoberts @George_Thompson 

[George_Thompson]

This is something that I would work with technical support on. I do not have any visibility behind the scenes and how the size is calculated or why it is increasing in your case.

[troyf_PTC]

I have struggled with this ever since we started using ArcGIS Online back in 2018, but recently it seems like it has gotten worse over the past couple of years and sometimes fixes that worked before don't always work now.

(I already have a case open under another account for this issue with a specific feature service that is sitting at over 1GB of feature data and 59GB of attachments that refuses to shrink, and attempts to overwrite the feature service through ArcGIS Pro were unsuccessful, but enough about that).

I always noticed this with layers that were updated via FME Desktop or FME Server (which just use the REST APIs behind the scenes). Usually I can eventually trim it down by removing all editing tracking, disabling offline access, and within the REST Admin page removing all replicas, removing and trimming the change tracking, and compressing the feature service.

But like you said, you don't have offline or editing enabled. I suspect that they are building in and using a lot more change-tracking tables behind the scenes, even when the "Keep Track of changes to the data" is supposedly disabled within the Feature Service settings page... but I don't have much proof to support that theory...