Data
10,000 synthetic points, matching CSV, GeoPackage and provenance metadata. Use the same keys in every check.
data/municipal-assets.geojson ↓Synthetic data · Downloadable code · ES / EN
A lab to test your Open GIS migration
Follow 10,000 fictional municipal assets from a file to an API, a map service and a viewer. The package brings together data, configurations, exercises and local results so you can repeat the checks and understand each component’s limits.
No registration. Local use. Fictional data near Medellín; it does not represent real inventories, people or infrastructure.
One dataset, several usage contracts
01
Files and editing
GeoJSON / CSV / QGIS
Identifiers, attributes and coordinates you can inspect.
02
Shared data
PostGIS
Loading, spatial index, roles and acceptance queries.
03
Services
GeoServer / pygeoapi
WMS images, WFS features and OGC API collections.
04
Consumers and decisions
MapLibre / TCO
Local viewer, dependency inventory and budget.
You do not need every service installed to explore the example. Start with the file and viewer, then choose the pygeoapi CSV path or the container stack. The instructions explain prerequisites, local passwords and cleanup commands.
What is in the package
QGIS
A starter project for exploring assets, categories and forms. Review it in your QGIS version.
qgis/municipal-assets.qgs ↓PostGIS
Schema, constraints, permissions and initial load, plus acceptance and export queries.
sql/01-init.sql ↓GeoServer
SLD styling and publication script; WMS/WFS requests and a locally generated map response.
geoserver/assets.sld ↓pygeoapi
CSV and PostGIS configurations, pinned dependencies, OpenAPI generation and repeatable measurement.
pygeoapi/config-csv.yml ↓Inventory and budget
CSV worksheets for dependencies, acceptance and TCO. Supply your own figures and responsibilities before deciding.
worksheets/migration-audit.csv ↓The ZIP also contains a MapLibre viewer with local assets, data and measurement scripts, Compose, ES/EN instructions, licenses and the file manifest.
What ran and what still needs testing
Observed local execution
- GeoJSON → GeoPackage conversion: geometries, attributes, identifiers and spatial index checked with GDAL. 6 checks →
- pygeoapi 0.24.0 · CSV · 11 passing checks; pagination, filters, identifiers and error responses.
- GeoServer 2.28.5 · Shapefile · 8 passing checks; WMS/WFS, pagination, filtering and anonymous administration denied.
- 210 measured requests in total, with samples and environment published. The image below is an actual GetMap response.
Validation required in your environment
- The Docker/PostGIS stack was not executed in this environment. Verify startup, SQL roles, queries and service connections.
- The QGIS project requires opening, visual review and editing checks. It is not presented as a QGIS screenshot or a QField test.
- Outside this run: concurrent users, TLS, corporate identity, high availability, ArcGIS clients and production data.
Local results, with the method visible
Each query had 5 warmup requests and 30 measured requests, one at a time, over local HTTP without TLS. Timing includes reading the complete response. p50 is the median; p95 uses nearest rank. Caches were not cleared and other development processes were active.
These figures describe one lab run. WMS returns images while WFS/OGC API returns features: these timings cannot rank products. They do not estimate production capacity or savings against ArcGIS.
pygeoapi · CSV
2026-09-23 · Darwin 25.6.0 · arm64 · Python 3.13.12 · Werkzeug 3.1.8
| Query | p50 (ms) | p95 (ms) | Response (bytes) | Errors |
|---|---|---|---|---|
| Page of 10 features | 37.55 | 58.65 | 3,311 | 0 / 30 |
| Page of 100 features | 37.32 | 53.17 | 23,498 | 0 / 30 |
| Spatial filter, limit 100 | 109.37 | 126.8 | 23,717 | 0 / 30 |
| District filter, limit 10 | 11.42 | 12.26 | 3,365 | 0 / 30 |
GeoServer · Shapefile
2026-09-23 · Darwin 25.6.0 · arm64 · Java 17.0.20 · -Xms128m -Xmx768m
| Query | p50 (ms) | p95 (ms) | Response (bytes) | Errors |
|---|---|---|---|---|
| WFS: 10 features | 316.94 | 452.7 | 2,678 | 0 / 30 |
| WFS: 100 features | 281.57 | 374.01 | 24,930 | 0 / 30 |
| WMS: 512 × 512 image | 23.26 | 29.16 | 57,175 | 0 / 30 |
An actual service response
GeoServer generated this 512 × 512 image through WMS 1.1.1 using the bundled style. The grid is deliberately synthetic. It is not real asset mapping or evidence of acceptance in another client.
Open original PNG →
From the exercise to your real workflow
- 1. Repeat the example without changing the data. If it fails, record version, request, response and the difference from the expected result.
- 2. Use an authorized, bounded copy of your data. Compare identifiers, geometries, attributes, rules and consumers.
- 3. Add permissions, restoration, volume and editing checks where your workflow requires them. A successful GET does not validate a complete migration.
- 4. Complete the inventory and budget. Preserve a rollback path and the ArcGIS components you still need.
- QGIS for ArcGIS users: your first reproducible project →
- PostGIS for cadastre: a minimal model, geometry and quality controls →
- Publish a PostGIS layer with GeoServer and verify WMS and WFS →
- pygeoapi: publish your first OGC API with test data →
- MapLibre: build a web map with GeoJSON →
- A phased ArcGIS to Open GIS migration plan →