General aspects

[iferencik]

GOAL -> semiautomatic geospatial assessment
- simple/intuitive
- fast
- store data cloud
- intermediary layers
- out layers
- avoid using connstring
- compute in cloud
- network speed
- scaling

	- authentication

PATH -> mask->admin->assessment->cloud storage

1. Simple/intuitive

   • minimize the number if interactions:
     input: raster/vector mask (bbox)
     oputput
     GPKG where every subcommand data is stored in tables starting with that subcommand
     tanzania.gpkg
     admin
     natenv.landuse, vegetation

              		population (male, female, dep...)
              		builtenv,(roads, elgrid, buildings)
              		socec (hdi, GNIU, GDP, etc)
              		
      intermediary
     

2. Fast/stable

   speed (parallelization) GDAL/RASTERIO
   stability
   windowed approach
   max mem
   keep state to recover

3.Storage
blob
fileshare
conn string should be avoided (azure identification lib)

4. Cloud compute

   1. install lib locally
   2. run it

   ++++++++++
   WE
   1. create image
   2. deploy to ACI
   USER
   3. runs command
   authenticates
   connects to ACI
   runs the command