Publishing a .tif image in arcpro as a "tile" vs "map image" layer type

[DamonNelton]

To start we have an enterprise base deployment with no additional add-ons such as image server.

We geo-reference drawings and images regularly and then publish them as a "map image" using the "share as web layer" to our portal/server. It seems to work in arcpro will little headaches. However when we do this the CPU, postgres, and arcsoc processes spike and bottleneck our enter enterprise system, sometimes so bad that it does not recover and we have to restart the server. I'm considering switching our workflow to publish as a "tile" layer type instead where we can cache locally. By doing this in theory I'll be putting the stress and processing power on my local machine instead of the enterprise server. Once done publishing and in our enterprise system they seem to have little impact on performance other then an additional service that refreshes however often we set it up for.

Currently all we use these .tif images for in portal is to view them overlayed in our maps as background images. 

Is there anything above that's not correct, and is there anything I'm not taking into consideration in regards to functionality differences between the two different layer types?

Thanks for any insight you can share

[Rafaelde_Oliveira_Casagrande · Accepted solution]

Hi Damon,

A Map image layer supports feature queryng and It draws dynnamicaly or from existing tiles. A Tile layer allows you fast visualization from existing tiles and It is published only in the hosting server (Tile cache data store).

After publishing large datasets caching on the server is common a peak in the CPU/memory.

For both services, to reduce the server consumption you can generate the cache locally (on your own desktop) in the tpk format and then upload to the server.

Following some links may help you:

Share a map image layer
Share a web tile layer

Best,
Rafael Casagrande