In some cases, uses MapProxy Python to deliver WMS, WMTS, TMS from the other servicesįor our GeoRequest Area of Interest API we considered Serverless but because the size of the source and bandwidth costs it did not make sense to host that in AWS or AZURE we went with a dedicated host at a colocation with 20TB of RAID 10 Storage and 32 cores for the jobs to run (and we can mail next day external hard drive to add new data products instead of FTP with 10 or 20 mbps upload) The Colocation is giving us 50TB of bandwidth a month for the cost of the dedicated server.
Removing the standard GeoServer/MapServer WMS Stack Or processing raster data to raster tiles and serving via tile server We've moved to using OGC WPS for some processingĪnd directly publishing data via POSTGIS with CrunchyData's pg_featureserver OGC API Features and pg_tileserver delivers dynamic PBF Vector TilesĪs well as processing data to vector tiles and then serving via tile server
The process must be short lived otherwise we've used serverless functions to start EC2/virtual machine and then run process and then stop EC2/VirtualMachine We've done AZURE functions and AWS Lambda Functions with our custom dotnet core console apps and calling GDAL, Whitebox, OTB and python Rasterio and other python packages. It can be used by Admins or given to end users depending on your needs. We have a CodeIgniter PHP Application that allows users to authenticate/authorize and upload GeoTIF or Shapefile and get WMS. Serverless would not deliver OGC WMS that would not make senseĪ better approach is just GeoServer REST API to publish data to coverage/data store and return a WMS, TMS, WMTS, WFS URL.
0 kommentar(er)
