bulk download distribution

It is access by artifact rather than interaction. The National Bridge Inventory's ASCII files record '~51 MB zipped for the combined all-states file, ~56 MB zipped including all records' under bulk delivery access; Natural Earth's Roads and Railroads ship 'Roads 8.66 MB zipped (v5.0.0)' plus a 'North America roads supplement 45.57 MB zipped (v4.0.0)'.

Overture Maps extends the model to cloud scale: its Transportation Theme notes that 'Data sits in public cloud buckets so there is no access fee' - distribution cost absorbed by the publisher, retrieval handled by the consumer's tooling.

Network datasets are heavy and query-poor. Routing, bridge-condition and corridor analysis need full graphs - pulling them record-by-record through an API would be impractical even where permitted, which is why every major transport geography here arrives as an archive.

Cost structure follows the delivery mode: with data in public buckets there is no access fee to negotiate, only egress and storage on your side. Version pinning becomes the buyer's job too - Natural Earth labels archives v5.0.0 and v4.0.0 explicitly, so pipelines should reference those versions rather than 'latest', or analyses will drift as baselines update.

1. Check archive sizing against scope. The ~51 MB NBI combined file covers all states; partial extracts should justify their smaller size.

1. Pin versions explicitly. Natural Earth's v5.0.0 and v4.0.0 archives show why versioned URLs matter for reproducibility.

1. Confirm bucket openness. Overture-style public buckets mean no fee but require your own sync tooling.

1. Match format to analysis. Zipped ASCII bridge files need parsing; GeoParquet-style cloud formats load directly into columnar engines.

• geoparquet-cloud-native-release - the modern cloud-bucket packaging these sources are moving toward.

Sector hub: highways-railtracks data.