FastVector VS NG911GISDataModel

You can use a package I created postgresVectorTileGenerator to seed tiles at larger zoom levels and load those into a cache sitting in front your tile service. The other thing you can is simplify your data or limit the number of features per tile. You can check out an API I created called FastVector. It allows you to limit the number of features per tile and dynamically cache data as well in a filestore. This will help at the higher zoom levels for larger datasets.

ArcGIS Online could be the right move as long as you set up the service correctly. You could also load the shapefile into a database like PostgreSQL and connect to the database via an API like Geoserver FastVector, or pg_tileserv. This would cause a lot more time, cost, and knowledge on your end though to host a database and api vs loading the data into the ESRI ecosystem.

Good idea, were you thinking of something like a cql_filter that I used in FastVector?

Here is an example code https://github.com/NENA911/NG911GISDataModel/blob/main/esri_geodatabase/NENA_NG911_Scripts/create_ng911_fgdb.py#L621 and the Toolbox is at https://github.com/NENA911/NG911GISDataModel/tree/main/esri_geodatabase

If you are looking to implement NENA NG9-1-1 fields, I would highly recommend that you take a look at the [NENA NG9-1-1 GIS Data Model Template](https://github.com/NENA911/NG911GISDataModel). It has recently been updated to NENA-STA-006.2-2022 and released on GitHub. It includes Python scripts and ArcGIS Toolboxes to created the v2 standard in both ArcGIS Pro and ArcGIS Desktop 10.x. Additionally, it includes SQL to create the v2 standard in PostGIS. There is extensive documentation in both versions.