Traditional GIS formats—File Geodatabase, Shapefiles, regular GeoTIFFs—were designed for desktop GIS in the 1990s. They don't work well in the cloud.
The Problem with Traditional Formats
Why File Geodatabase Fails in Cloud
- ✗ Must download entire file to read a single feature
- ✗ No HTTP range request support
- ✗ Locks prevent concurrent access
- ✗ Can't query from S3/Azure Blob Storage
- ✗ No integration with modern data platforms
Result: 10GB File Geodatabase takes 15 minutes to download just to read 1 feature.
Cloud-Optimized GeoTIFF (COG)
What: A regular GeoTIFF with internal tiling and overviews, optimized for HTTP range requests.
Why it matters: Read only the pixels you need, instantly, from cloud storage.
Performance: Regular GeoTIFF vs COG
Regular GeoTIFF (10GB)
14 min
Download entire file
Cloud-Optimized GeoTIFF
0.8 sec
Read 1 tile via HTTP range
GeoParquet
What: Apache Parquet with geospatial extensions. Columnar storage optimized for analytics.
Why it matters: 10-100x faster queries than Shapefile/FGDB. Native integration with Databricks, Snowflake, BigQuery.
STAC (SpatioTemporal Asset Catalog)
What: JSON specification for cataloging geospatial assets (imagery, rasters, point clouds).
Why it matters: Standardized search and discovery across petabytes of imagery.
Migration Path
- Raster data: Convert GeoTIFF → COG using GDAL
- Vector data: Convert Shapefile/FGDB → GeoParquet
- Imagery catalogs: Create STAC catalog for discovery
At Axis Spatial, we've migrated 50TB+ to cloud-native formats. The performance gains are immediate and dramatic.