ArcGIS Enterprise 11.0: How “data store” can allow “exporting data, and adding or deleting fields?

[JamalNUMAN]

ArcGIS Enterprise 11.0: How “data store” can allow “exporting data, and adding or deleting fields?

It’s mentioned in the manual that “hosted feature layers that rely on the relational data store allow you to more easily share and manage feature data in your organization” amongst “exporting data, and adding or deleting fields.

How this(“exporting data, and adding or deleting fields) can be performed?

https://enterprise.arcgis.com/en/portal/latest/administer/windows/what-is-arcgis-data-store.htm

[Scott_Tansley]

Through the Enterprise Portal, just like exporting etc. in ArcGIS Online.

[NathanEnge]

The ArcGIS datastore is used only as part of a federated Portal environment. The add/update/delete capabilities on an hosted feature layer are accessed via the Portal interface or its corresponding Rest API.

[JamalNUMAN]

For example, I have “parcel” layer that’s published as map image layer, them how fields of the attribute table can be added, deleted, updated, etc.?

[ChristopherCounsell]

Go to the Item Details Page > Data to manage it.

Or via the REST API.

It is equivalent to ArcGIS Online. You do everything via REST API or applications (ArcGIS Pro, ArcGIS Portal) that interface with it.

[Scott_Tansley]

If you're using a Map Image Layer, then by default you're using referenced data.  You make the changes in excatly the same way as you would in a file or enterprise geodatabase, and then republish the service to reflect the changes.  

Note that you need to stop the service to make the changes.

You only make those changes via the Enterprise Portal for Hosted Feature Layers.

[JamalNUMAN]

But if the referenced “map image” option is selected, then why do we need to re-publish the service to get the edits reflected?

With the regular “ArcGIS Server”, we don’t need to republish the data to get the edits reflected.

Why the “federated arcgis server” requires republishing the data to get the edits reflected?

[ChristopherCounsell]

The behaviour of a referenced service is more or less the same for non-federated and federated ArcGIS Servers. The difference is the security model: users authenticate through Portal for ArcGIS (opposed to Server). An item is also created for the services in Portal for ArcGIS, to allow the service to be shared using the Portal security model (users, groups etc).

It's probably tied to caching. When publishing a Map service, you can choose how the map will draw from the data. From the data every time or from a cache. I would check these settings:

https://pro.arcgis.com/en/pro-app/latest/help/sharing/overview/configure-a-map-service.htm

[JamalNUMAN]

Despite the disadvantage of feature layer service (published with relation arcgis data store) of having the edits performed in the source data not reflected directly in the service but needs overwrite publishing, the feature layer has the advantage that it behaves as if it’s a feature class and thus symbology, labeling, analysis, export, etc. are all possible

[NathanEnge]

Map Image layers (traditional stand-alone arcgis server map services) VS hosted feature layers (Federated ArcGIS Server to Portal [Enterprise]) is an on-going debate amongst GIS Users. 
Most of my customers have adopted an approach where they leverage both the Hosted feature services, as well as traditional map image layers - with the map image layers' data source originating from an enterprise geodatabase like SQL Server or Oracle, etc.

You can do both; however, be aware, from an architecture point-of-view, it is recommended that an ArcGIS Server designated as Hosting server should be a separate from another ArcGIS Server doing traditional map layers.