A breakdown of pgRouting and OSRM vs per-call routing APIs, with real numbers from production fleet deployments.
Per-call routing APIs are simple to integrate and painful to scale: a fleet making tens of thousands of routing calls a day can watch its mapping bill climb into a meaningful line item, with no way to reduce it besides reducing usage.
pgRouting and OSRM, self-hosted on infrastructure you already control, replace the per-call cost model with a fixed infrastructure cost. Once the road network graph is loaded and the routing engine tuned, the marginal cost of an additional route calculation is close to zero.
Across production fleet deployments we have run, self-hosted routing has cut mapping API costs by 78–95% versus the equivalent volume of per-call requests, with the added benefit that route data and business logic stay inside your own infrastructure rather than round-tripping to a third party.
The trade-off is upfront engineering: standing up and maintaining the routing engine, keeping the road graph current, and handling the operational load that a managed API used to absorb. For fleets past a certain call volume, that trade consistently pays for itself within months.
Why 80% of PIA A55 submissions fail, and how spatially accurate, specification-compliant documentation reduces rejection rates.
SurveyingInside our AI-accelerated pipeline: from raw point cloud ingestion to classified LAS delivery at 97.2% accuracy.
GIS EngineeringA practical guide to preparing spatial asset data for regulatory compliance: schema validation, CRS alignment, and submission-ready formatting.