USF-IMARS / wv-land-cover

:earth_americas: Processing scripts for decision-tree land use classification on worldview 2 imagery
5 stars 5 forks source link

plan azure deployment #11

Closed 7yl4r closed 5 years ago

7yl4r commented 5 years ago

One aspect of this is to identify best way to run matlab scripts:

  1. automate USF licence setup (needs VPN setup too) VPN cost $30/mo -- I wonder if we could fake this w/ something like sshuttle
  2. use Azure "Data Science" VM that comes w/ matlab
  3. port the code into python
  4. shared matlab install over NFS
  5. (maybe) it will just work without concern...

The broader issue however is to plan how to deploy onto Azure. This includes storage & airflow networking as well as deployment of the VM images themselves.

puppetlabs/azure may be useful for deployment, but it might actually be easier to do things manually at our (small) scale.