Evaluating ArcMap to PostGIS migration for our org - seeking real-world experience

[Nabil007]

Our team is being forced off ArcMap by the 2026 retirement deadline and exploring whether to move to ArcGIS Pro or migrate select datasets to PostGIS/Postgres.

Before we commit engineering resources to either path, trying to understand the actual migration effort for complex File Geodatabases (relationship classes, domains, geometric networks).

Specific questions:
1. Has anyone here successfully migrated production FGDBs to PostGIS? Was the schema conversion or the data migration harder?
2. For those who tried ogr2ogr, how did you handle relationship classes that don't map 1:1 to Postgres foreign keys?
3. If you outsourced this work, what did it cost per GB (or per table)?

We're considering building internal tooling for this if the existing options (FME, manual) prove too expensive for our 200+ table migration.

Not selling anything - genuinely trying to scope the project.

[George_Thompson]

I have a few questions:

1. Are you still planning on using ArcGIS Pro for editing the data if you move to PostgreSQL?
2. Do you have multiple editors of the data at the same time?
3. Are you publishing out the data to services of any kind?
4. Is ArcGIS Enterprise (or Server) used at all?

Generically, you will lose the true functionality of the geodatabase if you move to PostgreSQL. This means that things you list above, plus other advanced geodatabase functions like topology / networks / feature datasets / / etc.

There are also some limitations if the data is in a database (with / without spatial type): https://pro.arcgis.com/en/pro-app/latest/help/data/databases/databases-and-arcgis.htm

Editing: https://pro.arcgis.com/en/pro-app/latest/help/data/databases/databases-and-arcgis.htm#ESRI_SECTION1_...

Hope this helps!

[Nabil007]

Thanks George—those are exactly the constraints we're weighing.

Context on our use case:

We're primarily doing analytical queries and API delivery (PostGIS + GeoServer/MapServer), not concurrent multi-editor workflows. The
geometric networks are largely legacy—we've already tested migrating those to PostGIS topology and pgRouting with acceptable performance hits.

The real blocker: FME licensing for 200+ tables with relationship classes gets quoted in
the "contact sales" tier, and ogr2ogr loses the domain/coded-value metadata that our downstream systems expect.

We're not trying to replicate ArcGIS Pro inside Postgres—just preserve data fidelity without the ArcObjects dependency.

Specific follow-up: For organizations that have gone the PostGIS route for read-heavy workloads, do you see performance
issues with PostgreSQL 15+ spatial types vs. enterprise geodatabases?

And has anyone benchmarked the bulk load time
differences (FGDB → PostGIS vs. FGDB → Enterprise GDB)?

If the consensus is "just bite the bullet and use ArcGIS Pro
with Enterprise," we'd rather know now before we commit engineering time to the migration tooling.

Side note: If anyone in this thread has already solved the relationship-class-to-FK mapping problem (especially for 1:M with attribution), would love to compare notes offline.